Ethereum Fusaka Upgrade: 8x Blob Capacity Reshapes L2 Scaling

Ethereum just shipped its second major network upgrade of 2026. Codenamed Fusaka, the upgrade introduces a fundamental shift in how the network handles data from layer 2 blockchains — and it could cut rollup costs to near zero. For builders shipping dapps on Arbitrum, Base, Optimism, and the dozens of other L2s, this is the infrastructure update that changes the unit economics of onchain applications.

The centerpiece of Fusaka is Peer Data Availability Sampling, or PeerDAS — a cryptographic technique that lets Ethereum nodes verify blob data without storing all of it. The result: an eightfold increase in blob capacity, with the first parameter bumps scheduled for December 2026 and January 2027. Here is what changed, how PeerDAS works under the hood, and what it means for the developers building the next generation of onchain apps.

What the Fusaka Upgrade Actually Changes

Fusaka is the first Ethereum upgrade to deliver on the network's rollup-centric scaling roadmap in a meaningful way. Since the Dencun upgrade introduced blobs in March 2024, Ethereum has supported a fixed ceiling of 6 target and 9 maximum blobs per block. That ceiling became a bottleneck as L2 adoption grew — blob fees occasionally spiked into the tens of dollars when demand surged, undermining the cost advantage of rollups.

Fusaka replaces that fixed ceiling with a scalable architecture. The key insight behind PeerDAS is that every Ethereum node does not need to store every blob. Instead, each node stores a random subset — initially one-eighth of the data — and uses erasure coding to reconstruct any missing pieces on demand. This means the network can handle far more blob data without raising hardware requirements for node operators, preserving Ethereum's decentralization while unlocking massive throughput.

As Alex Stokes of the Ethereum Foundation put it during the upgrade livestream: "It lets us scale while not compromising on the values that are so important to Ethereum."

How PeerDAS Works Under the Hood

PeerDAS is specified in EIP-7594 and represents years of research into data availability sampling. The core idea borrows from techniques used in peer-to-peer file sharing networks, adapted for the stricter security requirements of a blockchain consensus layer.

Here is a simplified breakdown of the flow: When a block is proposed, the blob data that L2s submit is erasure-coded — a mathematical transform that expands the data and enables reconstruction from any sufficient subset. Each validator node then downloads only a fraction of the coded data, typically one-eighth in the initial configuration. Before attesting to the block, the node performs random spot-checks against other peers to verify the full data is available across the network. As long as enough honest nodes each hold a piece, any missing segment can be reconstructed.

The cryptographic guarantees are remarkably strong. The probability of missing data slipping past the sampling mechanism is cryptographically negligible — between one in 10^20 and one in 10^24, according to the Ethereum Foundation's documentation. For context, that is roughly the probability of guessing a random 256-bit private key on the first try.

This architecture decouples blob throughput from node hardware. Before Fusaka, increasing the blob count meant every node had to download and store more data — a direct tradeoff against decentralization. After Fusaka, blob counts can scale independently of node requirements, which is why the theoretical ceiling jumps by a factor of eight immediately and could reach 16x or 32x in future upgrades as the storage ratio is reduced further.

The Blob Scaling Roadmap: BPO1 and BPO2

The Ethereum core developers are not flipping the switch to full 8x capacity overnight. Instead, Fusaka enables a new mechanism called Blob Parameter Only (BPO) forks — lightweight parameter adjustments that do not require a full network upgrade. Two BPOs are already scheduled:

BPO1 is scheduled for December 9, 2026 and will raise the per-block blob target from 6 to 10, with the maximum rising from 9 to 15. BPO2 follows on January 7, 2027, pushing the target to 14 and the maximum to 21 blobs per block. At that point, the network will have moved from handling roughly 1.125 MB of blob data per block to approximately 2.625 MB — more than a 2.3x increase in just two months.

The gradual approach is intentional. "Given this is a very new technique, and we are not sure how the network will respond, this is not the wisest decision," Stokes explained, referring to the option of dialing capacity to maximum immediately. The staged rollout lets the community monitor network stability, validator participation rates, and blob fee markets at each increment before moving to the next.

Longer term, the storage ratio per node could drop to one-sixteenth or even one-thirty-second of each blob, enabling proportional throughput increases. Ethereum co-founder Vitalik Buterin also noted on the livestream that blobs may eventually carry layer 1 calldata as well: "We think of blobs as being for L2s. In the long term, we want to dump L1 data into blobs as well."

What Fusaka Means for L2 Builders and Users

For developers building on Ethereum L2s, Fusaka directly improves the economics of their applications. Lower blob costs mean cheaper transactions for users, which in turn enables use cases that were previously marginal — high-frequency DeFi strategies, onchain gaming with real-time state updates, social applications with frequent low-value interactions.

The upgrade also strengthens the argument for building on Ethereum's rollup ecosystem rather than alternative L1s that sacrifice decentralization for throughput. By solving the data availability bottleneck at the protocol level, Ethereum can match or exceed the throughput of monolithic chains while maintaining the security guarantees of its decentralized validator set.

Paul Brody of the Enterprise Ethereum Alliance captured the long-term vision on the livestream: "We actually already have more network transaction capacity than we really need. But we are laying the foundation on the road to a trillion transactions a day, and we will need every single upgrade by the time we are done."

For teams building consumer-facing dapps, infrastructure improvements like Fusaka remove friction from the onboarding experience. Users no longer need to understand gas markets to interact with an application — costs drop low enough that protocols can subsidize transactions or build gasless experiences without breaking their unit economics.

If you are building on Ethereum L2s and want to ship faster, thirdweb offers developer plans that scale with your project — from smart contract deployment to frontend SDKs across all major EVM chains. The infrastructure is getting faster and cheaper. The tooling should too.

Beyond Blobs: Biometric Signing and DoS Protection

While PeerDAS dominates the headlines, Fusaka ships several other improvements worth noting. The upgrade adds support for biometric transaction signing, letting users authenticate transactions using smartphone face recognition or fingerprint sensors — a step toward making self-custody accessible to a mainstream audience that finds seed phrases intimidating.

Fusaka also hardens Ethereum against denial-of-service attacks by improving how the network handles transaction spam. These defenses operate at the networking layer and make it more expensive for attackers to flood validators with malformed or redundant data, reducing the attack surface for a class of exploits that has occasionally disrupted block production on other chains.

What Is Next: The Road to Glamsterdam

Fusaka is not the finish line. Ethereum's next named upgrade, Glamsterdam, is expected to ship by mid-2026 (note: this timeline may shift). Glamsterdam will include improvements aimed at reducing the cost of using Ethereum mainnet directly, building on the foundation Fusaka lays for data availability scaling.

The pace of upgrades is also accelerating. Fusaka shipped just seven months after Pectra, which activated in early 2026. This reflects a broader shift in Ethereum's development culture — a renewed urgency driven by a crisis of confidence earlier this year and increasing competition from alternative blockchains. The message from core developers is clear: Ethereum is no longer moving at a once-per-year cadence.

For builders, the takeaway is straightforward. The cost of running applications on Ethereum's ecosystem is on a sustained downward trajectory. Every upgrade cycle brings cheaper L2 transactions, wider data pipes, and better developer primitives. The infrastructure is maturing, and the window for building category-defining onchain applications has never been wider.