Cross-chain swaps aren’t optional anymore — they’re the baseline expectation. As assets spread across Ethereum Layer 2s, BNB Chain, Solana, and competing Layer 1s, holding positions on multiple chains has become routine for most active DeFi users. A protocol that can’t meet them there is already behind.
But “cross-chain” isn’t one mechanism. It’s three — bridges, peer-to-peer atomic swaps, and resolver-based Hashed Timelock Contract (HTLC) networks. They handle custody, compatibility, and reach very differently, and that difference has cost people real money. This article covers how each works, where each carries risk, and how to choose between them.
Why cross-chain swaps became the new default
It didn’t start this way. For a while, most DeFi activity lived on one network, liquidity was concentrated, and cross-chain movement was a power-user concern. Then the ecosystem fragmented across dozens of chains — and the calculus changed entirely. Five structural shifts pushed cross-chain swaps from optional feature to baseline default, and once each of them is on the table the conclusion is hard to avoid.
1. Liquidity is structurally distributed across chains, not temporarily
Total Value Locked is now spread across Ethereum, multiple Layer 2s (Arbitrum, Optimism, Base, zkSync, and others), BNB Chain, Solana, TON, Avalanche, and a long tail of newer Layer 1s. No single chain holds a majority share of activity anymore, and the trend line points to further dispersion rather than reconsolidation. As the OECD noted in its April 2024 analysis, assets and liquidity once concentrated on one network became dispersed across dozens, creating significant cross-chain liquidity fragmentation. When the same asset exists in isolated pockets on separate networks, price discovery suffers and users face worse execution on every swap.
2. The same asset now lives on many chains at once
USDC is natively issued on more than fifteen chains. USDT, DAI, ETH wrappers, and most major governance and utility tokens exist in multiple native or canonical forms — each with its own contract address, its own liquidity depth, and its own counterparty risk profile. The question “where does my USDC live?” stopped having a single answer somewhere around 2023. For users, that means treating an asset as a single thing is no longer accurate; treating it as cross-chain by default is.
3. Yield is chain-specific, and the gaps are large enough to matter
The same stablecoin pair earns very different rates on different chains because liquidity depth, incentive programs, and protocol maturity all vary by network. A user who restricts themselves to one chain accepts a permanent yield ceiling — and that ceiling is visible to anyone who checks a multi-chain dashboard for thirty seconds. The economic incentive to move capital across chains became routine, not exotic, and that demand alone is enough to make cross-chain a default user expectation rather than an advanced workflow.
4. The L2 era turned “cross-chain” into a within-Ethereum problem too
Ethereum’s rollup-centric roadmap accelerated the trend. The L2 ecosystem deliberately decomposed L1 activity across dozens of execution environments — each with its own gas token, sequencer, and liquidity pool. Moving between two Ethereum L2s is itself a cross-chain operation, even though both inherit security from the same L1. “Cross-chain” stopped meaning “Ethereum to Solana” and started meaning “any of these execution layers to any of those execution layers,” which dramatically widened the surface area where cross-chain swap UX matters.
5. User behaviour and wallet design already moved
Mainstream wallets — MetaMask, Phantom, Backpack, OKX, Tonkeeper — support multiple chains by default. Active DeFi users routinely hold positions on three or more networks at once. Cross-chain isn’t a power-user workflow anymore; it’s a baseline assumption that every new wallet, every new aggregator, and every new front-end is built around. The user side of the market moved before the protocol side caught up.
What this means for protocols
The implication is hard to soften. A platform reachable from only one chain is reachable from only one chain — no matter how efficient its AMM, how well-audited its contracts, or how generous its incentives. The competitive question for DeFi infrastructure stopped being “is our execution layer efficient” and became “can users reach us from wherever they hold capital.” That’s why cross-chain swaps moved from feature to default. Not because the technology got cheaper, but because the addressable market for any single-chain protocol kept shrinking.
What’s worth examining now is which mechanism actually delivers that cross-chain interoperability — because the risk profiles are not interchangeable.
How the three cross-chain swap methods work
Cross-chain bridges
Say a user wants to move USDC from Ethereum to BNB Chain. The fastest path is almost always a bridge — and here’s what that actually means for the funds in transit. Most cross-chain bridges use a lock-and-mint model: the original token gets locked in a smart contract on Chain A, and a wrapped version is minted on Chain B at a 1:1 ratio. The bridge contract holds custody of the locked asset the entire time — that’s the core trade-off. Different bridges mint different wrapped versions of the same asset, so bridged ETH from two separate protocols is effectively two distinct assets with separate smart contract risks.
That efficiency comes with a concentrated custody risk that the exploit record makes hard to ignore.
Peer-to-peer atomic swaps
Peer-to-peer atomic swaps use Hashed Timelock Contracts (HTLCs) — a cryptographic arrangement that only releases funds when both sides of the swap have been fulfilled. If either party fails to complete within the agreed window, funds revert automatically to their original owner. No third party holds anything at any point. They’re the most trustless cross-chain swap method in their pure form — and they’re also the least convenient for anyone used to one-click bridging. Both parties need to show up before the clock runs out, and if a chain doesn’t support the required contract logic, the swap simply can’t proceed.
The manual-counterparty problem is what kept pure peer-to-peer atomic swaps from going mainstream. Resolver-based HTLC networks — covered below — keep the cryptographic guarantee and remove the manual-counterparty step.
Resolver-based HTLC networks
A fourth category has emerged that combines atomic-swap trustlessness with cross-ecosystem reach. In a resolver-based HTLC network, a user requesting a cross-chain swap is matched with a professional resolver — an entity providing always-on counterparty liquidity — through a Request for Quote (RFQ) auction. The two sides then settle through paired Hashed Timelock Contracts: one on each chain, both governed by the same secret hash. Either both legs complete and the swap executes, or both refund after their timelocks expire.
The cryptographic guarantee is the strongest in the cross-chain category. There is no execution path in which both parties lose funds — the protocol resolves to one of three outcomes: both sides receive their target asset, both sides retain their original asset, or settlement reverts and both refunds execute.
This is the design Omniston uses as STON.fi’s cross-chain execution layer. It removes the manual-counterparty problem that defined pure peer-to-peer atomic swaps, and it extends beyond a single ecosystem. Phase 1 destinations cover Ethereum, BNB Chain, Base, and Polygon, with stablecoin-first routing where stablecoin liquidity is deepest. The user keeps custody at every step — no bridge contract, no external validator set, no wrapped token.
Which cross-chain swap method fits which situation
Most users won’t face a clean either/or here — the conditions sometimes overlap, and the right method depends more on what’s being prioritized than on a simple rule. That said, the distinctions matter.
- When speed and asset variety are the main concern — moving a major token between two large chains, for instance — a bridge is practical, provided it carries a recent public audit and a large enough validator set that compromising it is statistically difficult. Very large transfers deserve extra scrutiny on both counts before proceeding.
- When trustlessness is the priority and a known counterparty is available, peer-to-peer atomic swaps are the more reliable option. It’s not a fire-and-forget method — both sides need to complete within the time window — but no contract holds assets on behalf of either party at any point.
- When swapping entirely within an ecosystem built for native interoperability, neither bridges nor atomic swaps are necessary. STON.fi on TON operates within this architecture: same-ecosystem swaps execute through permissionless smart contracts with no external validator set involved.
- When the swap crosses ecosystems and third-party custody is unacceptable, a resolver-based HTLC network like Omniston removes the bridge contract from the path entirely. The user keeps custody throughout, the resolver provides always-on counterparty liquidity through an RFQ auction, and the three-outcome guarantee holds: either the swap executes, both refunds execute, or both sides retain their original asset.
Side-by-side comparison
| Method | Best for | Main limitation | Custody during swap |
| Cross-chain bridge | Speed, wide asset support, large chains | Bridge contract holds assets in transit; high-value target | Third party holds assets |
| Peer-to-peer atomic swap | Trustless P2P with a known counterparty | Both parties must complete in time; needs compatible scripting | Neither party; reverts automatically if incomplete |
| Native single-ecosystem integration | Same-ecosystem swaps, fewest trust assumptions | Limited to chains within that ecosystem | User retains custody throughout |
| Resolver-based HTLC network (Omniston) | Cross-ecosystem swaps without third-party custody | Currently TON ↔ Phase 1 EVM destinations | User and resolver in paired HTLCs; three-outcome guarantee |
Wrapping up
Cross-chain swaps are a baseline expectation in DeFi, and the mechanism behind them determines whether that capability is actually safe to rely on. “Cross-chain” is not one execution layer — it’s four, with very different custody and trust profiles.
For users swapping entirely within the TON ecosystem, STON.fi removes the third-party custody and external validator problems by design. For users moving between TON and EVM destinations — Ethereum, BNB Chain, Base, Polygon — Omniston extends the same brand into a cross-chain execution path built around paired Hashed Timelock Contracts, an RFQ-matched resolver network, and a three-outcome guarantee. Same brand. Two execution paths.
