Cross-Chain Arbitrage & Bridge MEV

You finished the on-chain arbitrage course knowing one beautiful guarantee: your whole strategy lives or dies as a single atomic transaction. Borrow, trade three pools, repay the flash loan — all of it lands together or none of it does. That atomicity is what let you run an arb with almost no capital and almost no risk: if the numbers didn’t work, the bundle just reverted and you walked away owing nothing but gas.

This course is what happens when you take that guarantee away.

The moment your arb spans two different chains — buy ETH cheap on Arbitrum, sell it dear on Base — there is no single transaction that can touch both ledgers at once. Each chain has its own blocks, its own validators, its own clock. Settlement splits into two legs separated by a bridge, and into that gap rush every risk atomicity used to hide: you hold inventory you can’t instantly close, you wait for finality while the price drifts, you trust a bridge that might get hacked, and you discover that “the prices were different on two chains” is a far more dangerous sentence than it was on one.

This sits at the very top of the crypto-quant ladder. It assumes you can already price a cross-DEX spread, size an arb against pool depth, and read the cost-stack-and-auction economics that bid most of the profit away. It assumes you know what a stablecoin is and how Ethereum’s rollup ecosystem is laid out. We take all of that and ask the harder question: what is arbitrage worth when settlement is no longer atomic?

The course builds the multi-chain picture, leg by leg:

• From atomic to at-risk — why crossing chains breaks the single-transaction guarantee, and why cross-chain arb is really the non-atomic, inventory-heavy cousin of CEX–DEX arbitrage, not pure atomic arb.
• How bridges move value — lock-and-mint, burn-and-mint, and liquidity-network (“inventory”) bridges; the messaging layer underneath; and why every design is fundamentally two transactions on two ledgers.
• The risks atomicity hid — inventory risk, bridge latency, and probabilistic finality / reorg risk, and how the exposure window between the legs is where the price gets to move against you.
• Inventory & rebalancing — why you must pre-position capital on both sides, how same-direction flow drains one inventory toward zero, and why rebalancing cost and idle capital cap the strategy’s capacity.
• Bridge trust & tail risk — classifying bridges by trust model, the multi-hundred-million-dollar bridge-hack history (Ronin, Wormhole, Nomad), and pricing a fat-tailed loss that can erase years of spread in one exploit.
• Shared sequencers & cross-domain MEV — where MEV reappears in a multi-chain world, what shared and centralized sequencers change, and the bonded-relayer, escrow, and intent-based designs that try to approximate the atomicity you lost.

By the end, “buy low on one chain, sell high on another” stops being a free lunch and becomes what it really is: a capital-intensive, latency-bound, tail-risky carry trade dressed up as arbitrage — one you can finally size, cost, and respect.