Why Shared Liquidity Pools Across L2s Are Economically Unstable

Fragmented state is fundamental. Each L2, like Arbitrum or Optimism, maintains an independent, non-synchronized state. A shared pool requires a single source of truth, which a decentralized network of sequencers cannot provide without introducing a centralized coordinator or a new consensus layer.

Incentives are fatally misaligned. Liquidity providers on a shared pool face cross-chain MEV risk and sequencer censorship risk without direct compensation. Protocols like Uniswap V3 concentrate liquidity based on local fee markets; a shared pool would dilute fee capture and create arbitrage opportunities for bots, not LPs.

The evidence is in the architecture. Projects attempting shared liquidity, like early Stargate models, rely on oracle-based rebalancing or LayerZero's Ultra Light Nodes, which are bridging abstractions, not true state unification. This creates latency and introduces new trust assumptions, defeating the purpose of a native, atomic pool.

Shared liquidity pools assume synchronous state, but L2s like Arbitrum and Optimism have asynchronous finality. This creates a latency arbitrage window where asset prices diverge between chains before LPs can rebalance.

Automated arbitrage bots dominate this window. Protocols like Across and Stargate finalize user transfers in minutes, but MEV searchers execute on-chain arbitrage in seconds. LPs consistently sell low and buy high.

The economic model is inverted. LPs provide capital and take risk, but arbitrageurs capture the value. This is a persistent negative-sum game for liquidity providers, making the shared pool model fundamentally unstable.

Evidence: Analyze any cross-chain DEX like a multi-chain Uniswap v3 pool. The LP's impermanent loss from inter-chain price divergence consistently outpaces fee revenue, a dynamic proven by the dominance of intent-based bridges like UniswapX which abstract liquidity sourcing away from static pools.

Shared liquidity is economically unstable because it decouples asset ownership from settlement finality. A user's deposit on Arbitrum is not the same asset as a withdrawal on Optimism; they are linked by a slow, non-atomic bridge like Across or Stargate.

Non-atomic settlement creates free options for arbitrageurs. When prices diverge between L2s, bots extract value from the shared pool before the bridging transaction finalizes, acting as a persistent liquidity tax on all users.

This is a first-principles arbitrage between settlement speed and capital efficiency. Protocols like UniswapX and CowSwap solve this for intents within a single domain by batching and settling atomically. Cross-domain liquidity pools cannot replicate this without a shared sequencer or a synchronous bridge.

Evidence: The 30-minute to 1-hour challenge window for optimistic rollup bridges is a direct measure of this instability. During this period, the bridged assets are not final, forcing the liquidity pool to either over-collateralize or accept the arbitrage risk.

The Bull Case is Capital Efficiency. A single liquidity pool spanning Arbitrum, Optimism, and Base eliminates redundant TVL. This creates deeper markets and lower slippage for users, mirroring the Uniswap V3 model but across chains. The vision is a unified liquidity layer for the modular stack.

The Fatal Assumption is Synchronized State. This model assumes all L2s are economically and temporally aligned. In reality, sequencer finality and bridge latency create windows where pool states diverge. A trade on Optimism uses stale price data from Arbitrum, creating guaranteed arbitrage.

Arbitrageurs become the real LPs. Protocols like Across and Circle's CCTP settle in minutes, not seconds. This delay turns shared pools into a free option for MEV bots. Liquidity providers face adverse selection, earning fees on losing trades and missing profitable ones.

Evidence from Existing Models. LayerZero's Stargate uses a form of shared liquidity but imposes message quotas and fees to manage cross-chain state risk. Its model proves that native asset bridging and localized pools (e.g., Aave V3) are more stable than a single shared balance sheet.