How Ethereum’s Fusaka Upgrade Is Changing Data Availability Pricing for Rollups

Ethereum’s Fusaka upgrade, activated on December 3, 2025, fundamentally restructures the economics of Layer-2 data availability. By introducing Peer Data Availability Sampling and raising the block gas limit from 45 million to 60 million units, the network increases transaction capacity while lowering the cost per byte for rollups. This shift moves Ethereum from a model where every node stores all blob data to one where nodes sample small chunks, reducing bandwidth demands by roughly 80% for full nodes. The result is a more scalable infrastructure that supports higher throughput without sacrificing decentralization or overburdening validators with excessive hardware requirements.

PeerDAS and the New Storage Model

Peer Data Availability Sampling, defined in EIP-7594, is the core mechanism driving these changes. Instead of downloading entire blobs, validators now verify data availability by sampling random portions from peers. Each full node stores only about one-eighth of the total blob data, yet the complete dataset remains reconstructable from any 50% of the network. This design allows Ethereum to scale blob space significantly without requiring linear increases in storage or bandwidth for individual operators. For solo stakers, disk usage and download bandwidth decrease by around 50%, though upload bandwidth may increase two to three-fold during activation as builders exchange sampling proofs. Large operators holding approximately 800 ETH face about 30% higher download bandwidth than pre-Fusaka levels but maintain manageable load with standard 100 Mbps upload capacity.

The transition also introduces a critical technical requirement for Layer-2 software. PeerDAS switches verification from blob proofs to cell proofs, meaning L2 systems must update their codebases or risk failed transactions after activation. While this imposes a short-term development burden, it enables long-term stability by preventing single points of failure in data propagation. The system relies on erasure coding to ensure that if enough randomly selected columns are available, the validator can be statistically confident that the entire dataset is intact. This probabilistic approach replaces the deterministic "download everything" model, creating a resilient network where data withholding is detectable and costly for malicious actors.

Flexible Scaling Through BPO Forks

Blob Parameter-Only forks, introduced via EIP-7892, allow Ethereum to adjust blob target and maximum counts without a full coordinated hardfork. This modular design enables faster responses to changing demand from Layer-2 networks. For example, blob targets could rise from 6 to 10 to 14, with maximums increasing from 9 to 15 to 21, all through lightweight updates. This flexibility prevents the bottleneck that occurred after Dencun, where fixed parameters limited growth despite rising rollup activity. By decoupling parameter adjustments from major releases, Ethereum can match supply to demand more precisely, avoiding periods of scarcity that drive up fees.

This mechanism works in tandem with the increased block gas limit set by EIP-7935. Raising the default limit to 60 million gas allows more transactions and smart contract executions per block when validators adopt the new client defaults. However, this is a client default rather than an automatic network-wide change on day one, meaning adoption depends on validator participation. The combination of higher gas limits and dynamic blob scaling creates a responsive environment where capacity expands as needed. Developers gain the ability to build data-intensive applications without worrying about rigid caps that might constrain their protocols during peak usage periods.

Stabilizing Fees with EIP-7918

Before Fusaka, blob fees often collapsed near zero during low-demand periods, creating unstable pricing signals for rollups. EIP-7918 addresses this by tying blob reserve prices to execution costs, preventing the fee market from dropping to 1 wei. This ensures healthy fee signals for Layer-2s by establishing a minimum baseline cost for data availability. The rule acts as a negative-feedback controller, keeping aggregate supply pinned at the target even when price levels fluctuate. By aligning blob pricing with execution fees, the upgrade prevents unfair arbitrage between blobs and traditional calldata, ensuring that data posting remains economically sustainable for all participants.

For rollups, this means predictable costs for posting data, which translates to consistent user fees. High-throughput chains that settle frequently benefit most because their continuous blob consumption becomes less volatile. The stable pricing model removes the incentive for inefficient usage behavior that plagued the early post-Dencun era. While some users might prefer lower absolute fees, the predictability provided by EIP-7918 offers greater value for developers planning long-term infrastructure. The question remains whether this floor will be high enough to discourage spam while remaining low enough to keep mainstream applications affordable as adoption grows.