How ether.fi Migrated $220M in Live Consumer Payments to OP Mainnet

The biggest migrations in finance are always disasters. TSB locked 1.9 million UK customers out of their own accounts for weeks after a botched platform switch in 2018. The outage made national news, triggered a parliamentary inquiry, and cost the bank over £330M in remediation. Santander, Barclays, NatWest have all had their own versions. The pattern repeats: announce the migration, take the system offline during a maintenance window, pray nothing breaks, spend weeks cleaning up when it does.

In crypto, the constraints are worse. There is no maintenance window. The infrastructure runs 24/7/365. You cannot tell a live payments product processing millions of dollars in daily transactions to hold tight while you swap out the engine underneath it.

So when ether.fi moved its entire product to OP Mainnet on April 15, cards kept working the whole time. $220M in DeFi TVL relocated in three days. The people tapping their ether.fi cards at a coffee shop in London or a gas station in Texas had no idea the infrastructure underneath their payment just changed.

What ether.fi Actually Did

ether.fi runs the largest non-custodial crypto card by spend volume. 70,000 active cards. 300,000 accounts. Millions of dollars in real-world payments every single day, backed by a $5.8B protocol.

The migration moved everything to OP Mainnet while the product was running at full volume. Old environment running, new environment coming online, traffic shifting over. Cards couldn't skip a beat.

A DeFi protocol can usually tolerate a brief pause. A yield vault going offline for a few hours is annoying but recoverable. ether.fi Cash has 70,000 active cards being tapped in real time. A declined card is a user experience failure. A pattern of them is a product trust failure. There is no "we'll be back in a few hours."

Who Noticed

You only hear about migrations when they fail. TSB made headlines for weeks. A clean migration is invisible.

ether.fi built custom monitoring specifically for the migration. Not a single unexpected alert triggered during the entire move.

This one got noticed anyway, though users never knew it happened. Institutional observers who run compliance desks and treasury operations at scale saw $220M move cleanly and flagged it as remarkable: no incident report, no customer communication, no "we apologize for the inconvenience."

The most common reaction: "wait, how did that actually work?"

How It Actually Worked

The Architecture

ether.fi's Scroll deployment stayed live while OP Mainnet stood up alongside it. There was no cutover moment. Core contracts deployed on OP Mainnet first, mirroring the exact system and ownership configuration running on Scroll. Then the peripheral layer: tooling to migrate individual user accounts across chains. During the build-out, new users were already provisioned with OP Mainnet accounts directly.

The key building blocks already existed: cross-chain bridging, deterministic vault deployments, card accounting that works independently from onchain settlement. Each component could be upgraded on its own. That sequencing meant no single step was all-or-nothing.

The old environment stayed live the entire time. If something had broken mid-migration, ether.fi could have kept routing through Scroll while the team investigated. The failsafe was built into the architecture from the start.

The Migration Stack

Three properties of OP Mainnet worked in ether.fi's favor. Gnosis Safe deploys deterministically across OP Stack chains, so ether.fi got the same multisig addresses on OP Mainnet without redeploying or reconciling address differences. The canonical bridge plus OFT support for weETH gave them bridging paths with known behavior. And the WETH predeploy handled ETH wrapping cleanly.

Three assets moved through bridges: USDC, USDT, and WETH each bridged from Scroll to Ethereum first, then took separate paths to OP Mainnet depending on the asset.

Every other asset moved as an OFT, bypassing Ethereum entirely.

On oracles, Pyth price feeds for EURC/USD, ETHFI/USD, and eUSD/USD were live on OP Mainnet before migration started, validated against Chainlink availability as a fallback. Optimism helped coordinate the timeline with Pyth directly. Push oracle model. No gaps in price data during the transition.

The Product

ether.fi paused deposits and withdrawals during the move but kept processing card payments the entire time. Their accounting system tracks every card spend independently from onchain settlement, so cards could keep authorizing while assets were in transit. Once migration completed, the team settled everything onchain on OP Mainnet and reopened deposits and withdrawals. At no point did a card decline that wouldn't have declined anyway.

ether.fi owned the migration end to end: contract deployments, engineering, migration tooling, user transition flows. OP Mainnet's permissionless infrastructure meant the core deployment required no special access or coordination. Optimism helped accelerate the timeline by coordinating third-party vendors, getting Pyth oracle feeds live and asset metadata confirmed on Etherscan before launch day.

The Performance Floor

OP Mainnet runs the specs a payments product requires: sub-250ms finality through Flashblocks, $0.00001 median transaction fees, 99.99% uptime, throughput at 20Mgas/sec scaling toward 100Mgas/sec, and data feeds powered by Pyth Network.

Spec sheets are easy to publish. The migration proved those numbers hold when a live product is moving $220M in real assets while processing consumer payments in real time.

What This Proves

ether.fi arrived on OP Mainnet with $220M in TVL. That number has already grown to $347M since the move. Gold Vaults, a Euro card, and native stablecoin support are next. ether.fi's roadmap starts here.

"OP Mainnet is the only place where the team that built the stack co-pilots your migration, and where the liquidity is already deep before you arrive. We closed the move in three days with zero downtime, and we're already building the next chapter."

Charles Mountain, DeFi Ecosystem Lead, ether.fi

Every protocol team sitting on infrastructure that isn't serving them already knows the migration calculus: the risk, the cost, the months of engineering time, and the fact that every example they can think of went badly.

ether.fi just rewrote them. Three days. $220M. Zero disruption. The playbook exists now. The next team doesn't have to write it from scratch.