Ethereum layer-2 networks need “responsive pricing” to scale to billions of users and reduce the fee swings that still accompany congestion, Offchain Labs co-founder Edward Felten said during a keynote at EthCC 2026.

Ethereum’s EIP-1559 upgrade launched in August 2021, as part of the London hard fork. It reformed the Ethereum fee market by modifying the gas fee limit and introduced a feature that burns part of the transaction fees, removing them permanently from circulation.

Felten said gas-price swings are still the main mechanism for protecting networks from being overrun during periods of heavy demand, even though that produces the kind of fee volatility mainstream users tend to reject.

“[With responsive pricing], you can see more traffic at lower gas prices without overrunning the infrastructure.”

Volatile gas prices have long been a barrier to mass adoption, particularly for users accustomed to fixed or predictable transaction costs in traditional financial systems.

The issue matters because Ethereum’s scaling story is no longer just about adding more throughput. It is increasingly about whether layer-2 networks can make transaction costs predictable enough for mainstream-style apps while still pricing congestion honestly enough to protect infrastructure under heavy demand. Arbitrum’s dynamic pricing rollout is now one of the first live tests of that tradeoff.

Arbitrum One the first L2 to adopt responsive pricing

Arbitrum One adopted dynamic pricing in January. It described the model as an “Arbitrum platform direction to make fees more predictable under demand by aligning prices with real network bottlenecks.”

Related: Gavin Wood’s biggest hope: Free crypto transactions and Web3 tech worldwide

Felten shared multiple charts showing how Arbitrum gas fees remained lower during peak network volumes than fees on the Base network and other L2s that rely on EIP-1559.

Arbitrum One is the largest L2 with $15.2 billion in TVL, while Coinbase’s Base Chain is second with $10.9 billion, according to data from L2beat. L2s are securing over $39.7 billion in cumulative TVL, up 4.6% over the past year.

While responsive pricing may be more scalable and more transparent about underlying costs, its biggest downside is lower predictability than EIP-1559, according to Julian Kors, a senior developer and founder of execution workspace startup Pulsar Spaces.

The debate is not about one model being better, but whether networks optimise for “predictability and mechanism design purity or for efficiency and real-time cost alignment. EIP-1559 does the first very well. Responsive pricing leans into the second,” he told Cointelegraph. 

Related: Ethereum Foundation accelerates 70,000 ETH staking plan after BitMine sale

Responsive pricing is a step forward, but the gas model needs replacing

Jerome de Tychey, president of Ethereum France and EthCC, told Cointelegraph that responsive pricing could improve user experience by making fees more closely reflect actual network demand.

Cyprien Grau, project lead at gasless Ethereum L2 Status Network, agreed, calling the new pricing model a “real improvement in fee accuracy.” However, the model still relies on a “fee market,” meaning that users may still face variable costs and gas spikes during congestion, he told Cointelegraph.

“It doesn’t solve the structural problem: L2 gas fees trend toward zero as scaling on L1 and L2s improves and competition intensifies. Responsive pricing makes the decline smoother, but you’re still building a revenue model on a depreciating asset.”

Grau added that responsive pricing is the “most advanced version of the gas model,” but said the gas model needs replacing. “L2s that scale to billions of users will be the ones where users never think about gas at all, and where networks’ economics don’t depend on charging them for it,” he added.

The fee model debate comes as parts of the Ethereum ecosystem are already rethinking the original rollup-centric scaling thesis. In February, Vitalik Buterin argued that some layer-2 assumptions no longer held and that future scaling should rely more heavily on the mainnet and native rollups.

L2 networks were created to scale Ethereum and offload part of the transaction load from the mainnet. However, Ethereum is now reconsidering its L2-centric approach, as these networks have siphoned significant economic value from the mainnet.

Magazine: Ethereum’s Fusaka fork explained for dummies — What the hell is PeerDAS?