Fusaka: Ethereum's evolutionary leap towards infinite scalability after Pectra

After the success of the Pectra upgrade, the Ethereum community is preparing for the next big step. On December 3, 2025, Fusaka will arrive, an hard fork embodying the vision of the network to achieve virtually unlimited scalability. The name itself reflects this ambition: Fusaka combines “Fulu” (execution layer) and “Osaka” (consensus layer), symbolizing the integration of the two pillars of the protocol.

Why Fusaka is crucial for the future of Layer 2

Recent years have shown how Layer 2 Rollups have become the main solution for high costs on Ethereum mainnet. However, these protocols still face significant obstacles: fees remain too high during congestion periods, and the network architecture is not yet optimized to handle massive data volumes. Fusaka directly addresses these issues through nine proposed (EIP) improvements, each designed to enhance a specific aspect of the network.

PeerDAS (EIP-7594): How Ethereum will verify data without overloading nodes

The introduction of EIP-4844 revolutionized data availability but created a new constraint: each node must download enormous amounts of blob data to verify its authenticity. This threatens network decentralization. Bandwidth requirements increase, decentralization levels decrease, and small validators struggle to keep up.

PeerDAS (Peer Data Availability Sampling) solves this dilemma by allowing nodes to verify data integrity by downloading only random fragments instead of the entire dataset. The mechanism works by dividing each blob into small units called “cells,” organized into columns. Each node is responsible for some specific columns and samples others from peers. If a node collects at least 50% of the total columns (for example, 32 out of 64), it can fully reconstruct the blob thanks to an erasure coding that adds redundancy to the data.

This approach creates a balance: validators, equipped with more powerful hardware, can store larger volumes and serve as network anchor points. Ordinary nodes remain active participants without bearing the full computational weight. The consequence? Ethereum can significantly increase blob capacity while keeping hardware requirements for participants low.

An important rule accompanies this innovation: no transaction can contain more than 6 blobs. This limit protects the system from abuse and better distributes the load across the network.

Gas repricing: MODEXP and security limits

Three EIPs address the delicate issue of gas pricing, each tackling specific problems in the MODEXP precompile mechanism.

EIP-7823: Putting a brake on MODEXP data

Ethereum’s precompile MODEXP has historically accepted input sizes that are theoretically unlimited. This has caused numerous consensus vulnerabilities: each client implemented the function differently, testing became impossible, and the pricing formula was unpredictable.

EIP-7823 introduces a simple but fundamental rule: base, exponent, and modulus cannot exceed 1024 bytes (8192 bits). This limit is safe for all practical applications—RSA cryptography uses keys up to 4096 bits, elliptic curves even less. Analyzing blockchain history from 2018 to January 2025, no successful MODEXP call ever exceeded 513 bytes. Therefore, this change does not invalidate historical transactions nor introduce new risks, but eliminates pathological cases that threatened network stability.

EIP-7825: The maximum gas cap per transaction

Another structural vulnerability: a single transaction can consume almost all available gas in a block (40 million). If someone sends a transaction with 38 million gas, the block becomes essentially unusable for other transactions, creating a denial-of-service-like effect.

EIP-7825 sets a strict limit of 16,777,216 gas (2²⁴) per transaction, regardless of the overall block limit. This ensures each block naturally contains more transactions, preventing a single operation from monopolizing the block. The choice of 2²⁴ is not arbitrary: it’s a power of 2 (easy to implement), large enough for complex contracts, and about half the typical block size.

The impact on the community is minimal—almost all current transactions consume far less than 16 million gas. Only rare extreme operations will need to be split into multiple steps.

EIP-7883: Recalculating the true cost of MODEXP

Historically, MODEXP operations have been underpriced relative to their actual computational cost. This creates a bottleneck: block producers process heavy calculations for small rewards, and attackers can fill blocks with costly operations without spending much.

Using an updated empirical formula, EIP-7883 increases the minimum cost from 200 to 500 gas and triples overall costs, with even higher penalties for operations with input over 32 bytes. The cost of operations on large numbers can increase by as much as 76-80 times. 99.69% of historical calls will see at least a tripling increase. This does not change how MODEXP functions but aligns the price with the actual work required.

Blob stability and proposers’ forecast

EIP-7918: Linking blob fees to execution cost

Blob fees (introduced by EIP-4844) fluctuate wildly. When execution gas dominates total cost for Rollups, lowering the base fee of blobs does not increase demand—a phenomenon called inelastic demand. The protocol keeps reducing the price down to 1 gwei (the absolute minimum), at which point the mechanism stops working.

EIP-7918 introduces a “reserve price” set as BLOB_BASE_COST × base_fee_per_gas ÷ GAS_PER_BLOB. This ensures that the blob base fee always maintains a sensible relation to execution cost, creating predictable stability for Rollups. An empirical analysis of four months of blockchain data confirms that the new mechanism prevents crashes at 1 gwei and drastically reduces volatility.

EIP-7917: Making proposer scheduling fully deterministic

The selection of proposer validators for future epochs is currently unpredictable. Even knowing the RANDAO seed, changes in actual balances (EB) during an epoch can alter the next epoch’s proposer list. This creates issues for pre-confirmation protocols and opens manipulation opportunities.

EIP-7917 solves this by introducing a deterministic mechanism that calculates and stores the proposer schedule for the next two full epochs at the start of each epoch. Once determined, the list no longer changes due to late EB updates. This predictability is essential for Layer 2 stability and prevents “balance brushing”—validators manipulating their balances after seeing the RANDAO.

Network security and efficiency

EIP-7934: A limit on block size

Without limits on RLP block size, an attacker can create enormous blocks that paralyze nodes and slow propagation. EIP-7934 sets the maximum at 10 MiB (with a 2 MiB safety margin), aligning with the existing gossip protocol limit in the consensus layer. This removes inconsistencies between layers and prevents DoS attacks based on excessive sizes.

EIP-7939: The CLZ opcode for fast bit operations

Developers have historically had to manually implement functions for counting leading zeros in Solidity, consuming excessive gas and bulky bytecode. EIP-7939 introduces a new native opcode CLZ (0x1e) at a cost of 5 gas, the same as ADD. This accelerates math libraries, compression algorithms, bitmaps, signature schemes, and cryptographic operations, reducing fees and zero-knowledge proof costs.

EIP-7951: Native support for modern hardware signatures

Apple Secure Enclave, Android Keystore, FIDO2/WebAuthn, and hardware security devices use the secp256r1 (P-256) curve. EIP-7951 introduces a precompile P256VERIFY at address 0x100, enabling Ethereum to verify ECDSA signatures on the P-256 curve securely and natively, at a cost of 6900 gas. This fixes security vulnerabilities of the previous proposal (RIP-7212) and finally allows users to access hardware-backed wallets with the same simplicity as Ethereum.

Conclusion: The scalable infrastructure of tomorrow

Fusaka is not a single revolutionary change but a coordinated wave of improvements addressing specific network constraints. PeerDAS enables data scalability, gas repricing ensures economic stability, proposer determinism enhances predictability, and new primitives optimize efficiency.

The result is an Ethereum ready for the future: Layer 2 Rollups can operate at lower costs and higher speeds, nodes remain decentralized thanks to sampling mechanisms, and network security is cemented by well-calibrated limits and incentives. When Fusaka activates on December 3, 2025, it will formally mark the transition toward the infinite scalability infrastructure Ethereum has always promised.