One of the not-so-great experiences in the Ethereum Ecosystem is being stuck waiting for the bridging to happen. Across is a permissionless bridge intents protocol (bridging vs cross-chain swaps) that fosters competition. They don’t want you to wait either, so they solved that exactly.

In this article, I use Across as an example to illustrate the efficiency of intent-based systems and discuss the journey of a protocol that has cleverly aligned people's incentives while allowing anyone to solve these intents in a permissionless manner, thus, providing an exceptional service to bridge users in the industry.

The Race to Fill Across Orders

Across is an intent based bridging protocol that supports bridging ETH, USDC and 10 other tokens between 6 EVM chains.

I do not want to do it. Because I have seen it a thousand times on X in the past 2 months. But my editor insists. So here we go. Yet another definition of intents.

Intents are any intentions a user would have that they are willing to pay others to fulfill.

Let's look at an example of bridge intent.

Imagine a user who wants to bridge 5 ETH from Arbitrum to Ethereum **for a small fee. It's a limit order. Here’s how it’s filled through Across:

1. The user deposits ETH on Arbitrum with a fee proposed by Across.

2. Many Across relayers race with each other to fill the order by making a transaction on Ethereum.

3. The winning relayer who fills the order gets paid on any chain they choose by Across after ~2 hours.

In Across' system, the relayers compete with each other to be the first to fill an order. The filler contract on the destination chain allows a unique pair of depositID and chainID to fill an order. After the first relayer fills the order, the second relayer’s transaction will inevitably fail. There is no second prize in relaying. So in essence:

Bridging is usually slow because the relayers have to wait for the source chain finality, otherwise, they risk losing funds. But Across’ bridge intents force the relayers to put a price on the finality risks in a race to fill the order. The key difference between any intent and a blockchain intent is that blockchain intents often have many fillers competing to realize users' intents.

Across’ focus on speed is evident in the chart below. Across has become faster with more competition over time.

Across relayers are like hawks, optimizing for the best node infrastructure to keep a close watch on the deposit transactions before they swoop down to make the destination transaction. If you're a solver looking to be an Across relayer, you now know what to focus on. Across is lucrative. I have a table in my Dune dashboard for Across that ranks the top Across relayers.

Across has won over a million impatient bridgoooor hearts by hosting these relayer races. But to offer the fastest bridging and best prices, Across needs to attract loads of relayers. And their strategy is simple.

A Low Barrier to Spin Up a Relayer

It is fairly simple to run a relayer bot for Across. That’s why around 29 relayers in the last one month were filling various Across orders. In comparison, UniswapX has 15 fillers. And Stargate just has one.

Note: The Across team runs the first 2 relayers. However, even then, 50% of the fees generated right now are going to other competing relayers.

If users had to rely on relayers from the Across team, it would have been a single point of failure, a trusted system, and a slow UX. In fact, Across had a single point of failure until the beginning of 2023. But they have realized that and have worked towards improving their system to attract more relayers to compete with them. The fact that it’s permissionless is enough to conclude that the users are getting market price and speed.

Other intent protocols will all be offering similar services like swapping, limit orders, TWAPs, bridging, etc., but the ones that win will have a robust solver network. So, keep your solvers happy, you need them. Your solver network is your moat.

One of the main tasks of a cross-chain relayer is to rebalance liquidity between chains. For example, UniswapX allows fillers to fill orders and accept users’ locked funds on the source chain. Rebalancing is a chore these fillers will have to handle themselves. But this creates an unfair advantage for fillers with fat portfolios, as they can rebalance in larger batches. We should wait and see how this plays out.

This is where Across is positioned better than UniswapX. Across uses its LP funds to help relayers get refunded on any chain they prefer, and the LP funds are rebalanced in batches. This allows smaller players with smaller pockets to come in and compete with faster fills. Across founder Hart L mentioned this on this X Spaces.

Everyone asks about the intents of the users, but nobody asks about the intents of the relayers. Except Across

There are consequences to offering this additional service to your relayers. Because, now you have just one rebalancer relied on by most of the relayers, a single point of failure. Maybe there are better ways to rebalance the funds.

Incentives to Innovate Around Rebalancing

To better understand Across’ advantage for incentivizing relayers, let’s take a closer look at how rebalancing works.

After the relayers fill user orders, they wait for 2 hours for UMA's optimistic oracle to verify the fills and refund the relayers on the chain they have chosen. If there isn't enough liquidity to refund on a specific chain, the locked Across LP funds are used to refund the relayers. To access these funds, an LP fee is deducted from the bridge amount.

But this posed a problem. As most relayers depended on Across' system to avoid rebalancing the funds, whenever there was an increase in bridge volume, there was also an increase in LP fees deducted from the users. Take a look at the spikes in LP fees (green line) prior to April 2023.

The relayers were not incentivized to request a refund on the source or any other chain. While the LP fee would increase during periods of high demand, which users would have to pay, the relayer fee would remain unchanged.

To address this issue, in April of 2023 (note that the LP fee never spiked after this timestamp), Across introduced the UBA fee model. This model incentivized relayers to receive refunds on a chain with higher liquidity by implementing a fee called the "balancer fee". As a result, the utilization of the LP funds decreased.

I believe LP funds will always be useful to batch the rebalancing of ETH from L2s to the Ethereum Mainnet. Therefore, the APY of the pooled funds will be equivalent to the interest rate for borrowing ETH. Across is essentially borrowing ETH for a duration of 7 days. This will always be an edge for Across.

Please refer to this page to understand this in a deeper manner.

In conclusion, Across just realigned the incentives and let the market forces take care of its problems.

Final Words

Across, as an intent-based bridge, is ready to blow LP-based bridges out of the water — as Across is more cost-efficient and lacks the large “honeypot” for attackers to target compared to liquidity networks like Stargate.

Furthermore, due to the permissionless nature of relayers and the decision to open-source relayer infrastructure, Across is far and away the best positioned to host a varied, secure solver set.

And, lastly, Across has built a rebalancing module that further incentivizes relayers to use their system.

Overall…

1. Low bridge fees

2. Super fast bridging

3. And batched rebalancing for their relayers

These features make Across a formidable player in the bridging industry. In the future, Across can enable Cross-Chain Swaps with some additions and removals to its existing architecture. Maybe just supporting many more tokens and removing the LP fee for accepting user funds on the source chain.

Hart said they are looking into it right now here 👀. The future is exciting.

I am grateful to have found extremely smart people like Arjun and Kram who reviewed this article and gave valuable feedback. Please DM if you’d like to review future articles. Maybe there’s a $SAND token on the way.

Sources:
My Across Dashboard
UniswapX kills Bridging
How Across Works

They were talking about single token bridging like from USDC on Optimism to Ethereum, and how Across and Connext architectures use an intent system to fill these orders. But my mind kept going to that one topic that wasn’t discussed, Cross-Chain Swaps.

“Cross-chain swaps through intent based protocols like UniswapX”.

I feel like this is a topic that hasn’t been touched upon by anyone. Surprisingly, even the intent protocols are not highlighting this feature. But, I think this can potentially change the nature of how users move assets between chains.

TLDR; Bridging is dead. Long live cross-chain swaps.

In this article, I will use my left-curve brain cells to explain how something like UniswapX will force existing bridges to evolve or go extinct.

Now, let’s go from sipping tea to sipping tech.

Bridging vs. Cross-Chain Swaps

All cross-chain swaps are bridging transactions, but all bridging transactions are not cross-chain swaps. Let me explain.

Warning: I will be using ‘token’ and ‘assets’ interchangeably, but they’re the same thing.

Bridging a token means getting the same token on a different chain.

This can be a wrapped asset through a lock/mint bridge, or a canonical asset through a liquidity or a burn/mint bridge. For example, bridging via Hop or CCTP to get USDC on a different chain is pure bridging.

On the other hand, cross-chain swaps means exchanging a token on one chain for another on a different chain.

This is a feature supported by only a select few, usually by combining bridges and DEXs in the same transaction. For example, Jumper Exchange would first bridge USDC to Arbitrum and then swap it to ARB using a DEX in a single transaction.

In my humble opinion, bridging is last cycle’s innovation — it gets the job done, but merely so. Cross-chain swaps are the evolution that’s just beginning to take shape and there are protocols that are already leading by example.

The Present State of Cross-Chain Swaps

Among existing protocols, Squid and Jumper are leading the parade. Squid supports cross-chain swaps through Axelar’s messaging layer and interestingly, Jumper can do the same with either Connext or Stargate’s messaging components (contract calls). But a cross-chain protocol like Bungee doesn’t allow one-click cross-chain swaps with destination swaps right now, you have to switch chains to sign the 2nd swap.

Let’s say a user wants to swap 1 ETH on Ethereum into 2000 ARB on Arbitrum. Here’s how Jumper would support this flow:

1. Bridge ETH to ETH on Arbitrum using Connext

2. Pass the destination swap callData through Connext

3. Swap ETH to ARB on Arbitrum using 1inch

Without Jumper’s cross-chain swap solution, this action could’ve taken 3 different interactions and a bridge delay to achieve the same result. This cross-chain solution offers a superior UX and is a notable innovation in the space.

Kudos to Jumper/LI.FI for being the inventor of one-click cross-chain swaps through Connext.

But a key inference here is that the destination swap happens after Connext’s bridging process. Say, this process takes 30 seconds, a lengthy duration in the liquid markets, in this time the price of ARB could’ve gone to the moon because some whale just bridged millions into Arbitrum. Or could’ve been nuked to oblivion because a whale bridged out of Arbitrum.

To avoid a lot of failures due to price volatility, Jumper has set the destination swap slippage to 1–2%. But…

Jumper trades-off on execution price for a superior UX.

Let’s check how significant the improvement is:

Say the price of ARB was 1 ETH = 2000 ARB when the transaction was signed with a 2% maxSlippage set, which makes the minAmount required for the swap to be successful as 1960 ARB.

This is how the transaction will play out for various scenarios:

So, the user never wins in a cross-chain swap through the existing systems. Because of these reasons, cross-chain swaps, although convenient, are quite inefficient on the existing solutions.

UniswapX — The Future of Cross-Chain Swaps?

L2s are gaining some notable traction lately, to the extent that it could be costly for Uniswap to lose users to platforms like Jumper as they offer a superior cross-chain UX. To get in on the cross-chain action, they’re launching UniswapX, an auction based trading protocol to support cross-chain swaps along with on-chain swaps. The timing couldn’t be more precise.

I believe UniswapX can change the status quo of the cross-chain landscape.

UniswapX’s cross-chain version is designed similar to their on-chain dutch auction system, where they auction off the order flow to the permissionless fillers to fill on the destination chain and receive funds on the source chain. Just like how the on-chain auction system will lead to better execution price because of filler competition, it will also lead to better cross-chain execution price mitigating the MEV risks involved due to the price volatility and the maxSlippage dynamic.

Let’s compare the execution of cross-chain swaps on UniswapX to the existing solutions with the help of an example.

Example: User wants to swap 1 ETH on Ethereum to 2000 ARB on Arbitrum.

UniswapX will execute the order as follows, from the perspective of the user:

1. Set the order parameters to start the auction at 2050 ARB, with a decay function to decrease the execution price by 10 ARB and a minimum price to be 1950.

2. The order would be sent to all the available fillers to bid on.

3. The execution price would gradually decay over time. Might go on for 3 mins.

4. A filler accepts to fill an order at price 2020 ARB and sends that amount to the user on the destination chain

If we run a simulation with various price changes, this is how UniswapX compares to the existing solutions in the space:

Clearly, UniswapX beats existing solutions in most scenarios (except when the slippage is exactly 2%). Pricewise, I’d choose UniswapX any day. But there’s a caveat here. There could be some trilemma here as well. Because UniswapX’s UX might not be as good as Jumper’s current UX. Again, a trade-off, a UX trade-off.

The Cross-Chain UX, Suffering from Success

There’s this misconception that UniswapX will drastically speed up a cross-chain transaction, thus, drastically. But this might not be entirely accurate.

First of all, UniswapX is a dutch auction system, where the price of execution gradually falls from a price above the market price until a filler agrees to fill the order. Because of this, there will be a delay from signing an order to getting the bids from the fillers. Currently UniswapX’s dutch auction goes on for 3 mins. After the dutch auction, the fillers would want to wait for a certain number of block confirmations for safety reasons.

Meanwhile, Jumper can fill an order in 27 seconds. Here’s an example of a cross-chain swap between the optimistic rollups where Connext destroys any bridge in comparison in terms of speed. Source tx and dest tx.

UniswapX can offer near-instant exits from an L2 to its parent L1 because of the instant confirmations you get on rollups (at least until they become real :wink:). However, on chains like Polygon that are notoriously known for reorgs, there’s no incentive for UniswapX Fillers to fill large orders as there’s a significant reorg risk.

If the fillers are filling a $100k order and the chain reorgs reverting the source chain transaction, the filler will have to beg for their money on-chain. Or maybe some validators can exploit this scenario. But, for small amounts, the fillers might choose to fill faster, although they don’t have any incentive to do so (no incentives for faster bridging). Maybe a user can specify a filler who then promises to fill faster for a higher fee. Maybe that’s one of their reasons to have the option to give a filler an exclusive right to fill your order.

This article I found by searching for “chain finality crypto” gives a decent overview of chain reorgs and how chain finality is interpreted and respected by various players like chain devs, CEXs, and bridges.

Summary: Comparison Table for Cross-chain Swaps

Final Words

UniswapX is prophesied to redefine cross-chain swaps, offering a superior alternative to existing platforms. While individual bridges might struggle to remain competitive, aggregators like Jumper could benefit by integrating UniswapX to enhance their offerings further.

I also recommend looking into DLN by deBridge which has a similar system to UniswapX, but it doesn’t involve Dutch auctions (Maybe it should, Alex). I call all the bridge builders out there to support dutch auctions for cross-chain intents/swaps.

I thank Arjun from LI.FI who entertained my thoughts and encouraged me to write about it.