Cross-chain trading at real, high volume throughput requires some centralization at the matching/clearing layer. In this article we talk about the tradeoffs needed for high throughput.
TL;DR
Cross-chain trading at real throughput requires some centralization at the matching/clearing layer at present blockchain scaling technology levels.
Centralized custody (CEXs) gives you speed and “all chains under one roof,” but you hand over the keys and take a single-entity failure risk.
Self-custody maximizes control, but true cross-chain trading devolves into bridges, wrappers, and OTC plumbing—and at scale you hit AMM slippage and on-chain throughput ceilings.
Decentralized MPC custody splits control across multiple independent entities. You keep non-repudiable control policies while tapping a high-throughput centralized matching engine for cross-chain execution.
Conclusion: For serious, scalable cross-chain trading, decentralized MPC custody may provide the best route. It’s the model Figure Markets is pioneering. Learn more in this report.
The Cross-Chain Reality Check
Everyone wants the same thing: trade any asset against any other—BTC vs. SOL vs. staked treasuries vs. tokenized equities—instantly and safely. Two hard facts stand in the way:
Throughput & latency: Public chains are improving, but matching millions of orders with microsecond-level determinism is a very different sport than producing a block every few hundred milliseconds. You need centralized matching to hit high-frequency numbers reliably, even if settlement is on-chain later.
Chain fragmentation: Liquidity, state, and asset registries live on different L1s/L2s. “Native” cross-chain isn’t a button you push; it’s an orchestration problem. Without a unifying layer, you end up hop-scotching through bridges, wrappers, and IOUs.
That’s why every workable cross-chain architecture centralizes something—the question is what you centralize and who controls movement of funds. That choice is your custody model.
Custody, Not Just a Wallet Setting
Custody is governance over movement. Who can move coins? Under what policy? With what redundancy? How is risk shared?
Below are the three models in the wild, using cross-chain trading as the lens.
1) Centralized Custody (Traditional CEX)
How it works
You deposit to a CEX (e.g., Coinbase). Assets sit in omnibus wallets the exchange controls.
The exchange abstracts away chains, maintains an internal ledger, and matches orders at very high speed.
To you, it feels like cross-chain: trade SOL/ARB/BTC/ETH pairs all day, instant fills, deep books.
Why it “works” for cross-chain
Throughput: The order book and risk engine are centralized, so matching and netting can run at HFT speeds.
Unified inventory: Because all deposits are pooled, the CEX can route fills across any listing without waiting for on-chain bridges.
Hidden tradeoffs
Single-entity failure: You’ve surrendered the keys to your coins and you’re opening yourself to risk of ops failure, hack, freeze, or policy change where your funds can be exposed.
Opaque rehypothecation risk: You can’t verify how your deposits are used internally.
Exit friction: High-stress events can turn withdrawals into a queue.
Bottom line: Great UX and speed. But the price of “seamless cross-chain” is trusting one entity with your coins.
2) Self-Custody (Pure On-Chain)
How it works
You hold the keys. All movement requires your signature.
Trading is done directly on DEXs, typically AMMs or on-chain order books on a single chain. Cross-chain requires bridges, wrappers, or OTC.
Why it struggles with cross-chain at scale
Bridges & wrappers: You’re juggling counterparty risk and contract risk across hops. Operationally messy, latency-heavy.
Throughput limits: Even fast L1s face mempool contention and block timing. On-chain order books degrade under bursty flow; AMMs fill but at price impact (slippage) when you size up.
Liquidity silos: You’re only as good as the pool depth on the chain you’re on—moving collateral across chains is slow or risky.
Yes, on-chain throughput keeps climbing. Solana leaders, for instance, have publicly framed NASDAQ-scale ambitions and even highlighted event-driven days where Solana DEX volume touched roughly 10% of NASDAQ’s daily volume—a remarkable milestone, but still episodic and not a daily, end-to-end replacement for centralized matching at sustained, institutional HFT scale
Bottom line: Maximum control, excellent for long-only, single-chain flow. For institutional-scale cross-chain trading, you usually end up with bridges, wrappers, and operational sprawl—and you still hit slippage and throughput walls.
3) Decentralized MPC Custody (The Middle Ground)
How it works
Your wallet keys are never whole. They’re split into cryptographic key shares held by independent operators (think regulated custody partners, governance nodes, HSMs/TEEs).
Movement of funds requires a threshold of these independent co-signers to jointly produce a signature (M-of-N). No single party, including the exchange, can move funds unilaterally.
On top sits a centralized matching & risk engine for speed. Orders are matched centrally; settlement occurs from the decentralized vaults per policy.
Why it unlocks cross-chain the right way
HFT-caliber throughput: Centralized matching delivers the latency you need for tight spreads, deep books, and risk controls across many markets.
No single-entity failure: Because keys are sharded across entities, one operator’s compromise doesn’t compromise the vault. Policy-based controls (whitelists, velocity limits, human-in-the-loop thresholds) add defense-in-depth.
Unified collateral view: You can cross-margin across chains without physically bridging every time. The custody layer authorizes movement atomically per policy; the engine handles netting and settlement cycles.
Auditability: You can prove how keys are governed, who co-signed, and under what policy—without re-introducing a single point of trust.
Practical tradeoffs
You accept centralized matching for performance. That’s the “some centralization” cross-chain demands.
Operational complexity: More moving parts: independent co-signers, policy engines, monitored HSMs/TEEs, and real-time controls. Worth it, because the payoff is resilience.
Bottom line: Decentralized MPC custody separates control of funds (distributed) from matching (centralized). That’s the winning architecture for safe, high-throughput cross-chain trading.
Learn more about MPC technology in this report.
Where Figure Markets Fits
Figure Markets is building around that decentralized custody core:
Keys split across independent co-signers so no single entity including Figure Markets can move funds on its own.
Policy-driven movement (amount limits, destinations, human approvals for large withdrawals) enforced at the custody layer, not just in app code.
Centralized, high-throughput matching so you can actually trade cross-chain at speed, with a unified view of collateral and risk.
Cross-asset netting and settlement designed to minimize bridges/wrappers while preserving control and auditability.
The result is the best of both worlds: CEX-grade execution with distributed custodial control.
The Punchline
Centralized custody gives you speed but concentrates risk.
Self-custody gives you control but fragments liquidity and throttles scale.
Decentralized MPC custody splits the atom: distributed control of funds + centralized matching for throughput = safe, scalable cross-chain trading.
That’s the design point Figure Markets is pioneering. If you care about living a world that enables trading any asset against any other, at speed, without trusting a single custodian with the keys, this is the path that actually scales. Learn more in this report.
Sources:
1: https://messari.io/report/state-of-solana-q1-2025
