Scaling is multi-dimensional. Throughput, cost, and latency improvements require separate, often incompatible, solutions. Optimistic rollups like Arbitrum reduce cost, but finality is slow. ZK-rollups like zkSync offer fast finality but face prover complexity. No single layer solves everything.
the-ethereum-roadmap-merge-surge-verge
Blog
Why Ethereum Scaling Comes in Small Steps
The Ethereum roadmap is a masterclass in incremental engineering. This analysis deconstructs the 'Surge' phase, explaining why its modular, rollup-centric approach—featuring Danksharding and EIP-4844—is the only viable path to global scale without sacrificing decentralization.
introduction
THE REALITY OF INCREMENTALISM
Introduction: The Scaling Delusion
Ethereum scaling is not a single breakthrough but a layered, iterative process constrained by security and decentralization.
The trilemma is a constant. Every scaling solution makes a trade-off. Sidechains sacrifice security for throughput. Validiums sacrifice data availability for cost. The market fragments into specialized chains like dYdX (app-chain) and Polygon zkEVM (general-purpose), each optimizing for a different corner of the trilemma.
Modularity is the only path. Monolithic scaling hits fundamental limits. The future is a modular stack: Ethereum for consensus/security, Celestia or EigenDA for data availability, and rollups for execution. This decomposition allows parallel innovation but introduces new coordination problems at the interoperability layer.
thesis-statement
THE ARCHITECTURAL CONSTRAINT
The Core Thesis: Modularity Demands Incrementalism
Ethereum's scaling evolution is a series of targeted, composable upgrades, not a single monolithic solution.
Monolithic scaling is impossible. Ethereum's security and decentralization are non-negotiable constraints. Increasing the base layer's throughput directly compromises these properties, as proven by high-TPS chains like Solana which trade decentralization for performance.
The modular stack forces specialization. Execution (Arbitrum, Optimism), data availability (Celestia, EigenDA), and settlement (Ethereum L1) become separate layers. Each layer optimizes for one function, creating a composable system of bottlenecks that must be solved sequentially.
Incrementalism enables permissionless innovation. Upgrades like EIP-4844 (protodanksharding) and danksharding target the data availability bottleneck first. This allows rollups like Arbitrum and StarkNet to scale before tackling the next constraint, the execution layer's proof verification speed.
Evidence: The roadmap is the proof. Post-Merge upgrades (Surge, Verge, Purge, Splurge) are discrete, non-breaking improvements. Vitalik's Endgame diagram depicts a multi-rollup ecosystem connected by shared security, not a single super-chain.
key-trends
WHY SCALING IS INCREMENTAL
The Three Phases of the Surge
Ethereum's scaling roadmap is a deliberate, multi-year crawl-walk-run strategy to preserve security while unlocking throughput.
01
The Problem: Monolithic Bottleneck
Pre-2022, Ethereum was a single chain executing everything. This created a fundamental trade-off: security and decentralization came at the cost of ~15 TPS and $50+ gas fees during peak demand.
- Congestion: Every app competed for the same global block space.
- High Cost: Simple swaps could cost more than the transaction value.
- Limited UX: Complex dApps (e.g., on-chain games) were economically impossible.
~15 TPS
Base Throughput
$50+
Peak Fee
02
The Solution: Rollup-Centric Roadmap
Ethereum's core devs, led by Vitalik Buterin, pivoted to a rollup-centric vision. Execution moves to Layer 2s (e.g., Arbitrum, Optimism, zkSync), while Ethereum Layer 1 acts as a secure settlement and data availability layer.
- Security Inheritance: L2s derive security from Ethereum's consensus.
- Scalability Leap: Aggregate throughput jumps to thousands of TPS.
- Cost Reduction: Fees drop by 10-100x versus native L1.
1000x+
TPS Potential
-90%
Avg. Cost
03
The Next Phase: Dank Sharding & Proto-Danksharding
The current bottleneck is data availability cost for rollups. EIP-4844 (Proto-Danksharding) introduces blob-carrying transactions, a dedicated data lane separate from execution. This is the precursor to full Danksharding.
- Blob Capacity: Targets ~1.3 MB of data per slot, a 10x+ increase in data bandwidth.
- Cost Predictability: Decouples L2 data costs from mainnet congestion.
- Enabler: Critical for scaling ZK-Rollups and high-throughput chains like Base and Starknet.
10x
Data Bandwidth
-100x
L2 Data Cost
DATA AVAILABILITY EVOLUTION
The Scaling Increment: From Proto-Danks to Full Danksharding
A phased comparison of Ethereum's data availability scaling roadmap, detailing the incremental rollout of Danksharding's core components.
| Feature / Metric | Proto-Danksharding (EIP-4844) | Full Danksharding (Target) | Pre-Danksharding Baseline |
|---|---|---|---|
Core Component | Blob-carrying Transactions | Data Availability Sampling (DAS) | Calldata |
Data Capacity per Block | ~0.75 MB (6 blobs) | ~1.3 MB (16 blobs) | ~0.09 MB (effective) |
Target Cost vs. Calldata | ~1% of calldata cost | < 0.1% of calldata cost | 100% (baseline) |
Data Persistence Window | ~18 days | ~18 days | Permanent (on-chain) |
Requires Consensus Change | |||
Enables Statelessness | |||
Key Enabler For | L2 fee reduction (Optimism, Arbitrum, zkSync) | Monolithic L2 scaling & high-throughput L1 apps | Basic smart contract execution |
deep-dive
THE CONSTRAINTS
Why 'Small Steps' Are Technically Non-Negotiable
Ethereum's scaling evolution is a sequence of constrained optimizations, not a single breakthrough, dictated by the blockchain trilemma and existing infrastructure.
Sequential Optimization is the only viable path. You cannot solve decentralization, security, and scalability simultaneously. Layer 2s like Arbitrum and Optimism first optimized for security via optimistic rollups, then for cost via data compression, and only now for decentralization with permissionless provers.
Legacy Infrastructure dictates the pace. Every major upgrade, from EIP-4844's blob space to EigenLayer's restaking, must integrate with billions in existing DeFi TVL and tooling from Chainlink or The Graph. A clean-slate design breaks more than it fixes.
Evidence: The transition from monolithic to modular execution, seen in the rise of rollups and Celestia, took five years. Each step required new cryptographic primitives (e.g., KZG commitments) and economic models that the ecosystem could absorb without systemic risk.
counter-argument
WHY INCREMENTALISM WINS
Steelman: The Case for Impatience
Ethereum's scaling roadmap is a marathon, but the market demands sprints. Here's why pragmatic, iterative solutions are dominating.
01
The Modular Dogma's Deployment Lag
Celestia, EigenDA, and Arbitrum Orbit promise a sovereign future, but launching a new rollup is a multi-month engineering effort. The market won't wait.
- Time-to-Market: ~6-12 months for a custom chain vs. ~1 week for an L2 deployment.
- Tooling Gap: Foundational infra like indexers, oracles, and wallets are not chain-agnostic.
6-12mo
Dev Time
~1wk
Alt Path
02
Blob Fee Volatility & The Appchain Tax
EIP-4844 (blobs) reduced costs, but didn't eliminate volatility. Dedicated blockspace is still a premium product.
- Cost Predictability: Appchains/Rollups pay a fixed overhead for security, avoiding the spot market's 100x fee spikes.
- Economic Model: Projects like dYdX and Aevo validate that predictable cost > absolute lowest cost for professional applications.
100x
Fee Spikes
Fixed
Overhead
03
L2s as the Ultimate Testnet
Optimism's Superchain, Arbitrum Orbit, and zkSync's Hyperchains are not the end-state. They are large-scale, live production experiments in governance, interoperability, and shared sequencing.
- Real-World Data: $30B+ TVL across major L2s provides a feedback loop no testnet can match.
- Iterative Sovereignty: Teams learn chain management on a managed L2 before graduating to a true rollup.
$30B+
Live TVL
Managed
Sovereignty
04
The Shared Sequencer Bottleneck
Espresso, Astria, and Radius are racing to decentralize sequencing, but today's L2s rely on a single sequencer for speed. Decentralization introduces latency.
- Performance Trade-off: Centralized sequencing enables ~2s finality; decentralized models target ~4-6s.
- Market Reality: Users and apps (e.g., perpetual DEXs) chose immediate speed over theoretical decentralization.
~2s
Current Finality
~4-6s
Decentralized Target
05
Interop is Still a Bridge Game
Native cross-rollup composability via shared proving or state proofs (e.g., zkBridge) is years out. Liquidity fragmentation is today's problem.
- Pragmatic Solution: Bridges like LayerZero, Axelar, and Wormhole move $1B+ weekly because they work now.
- Developer Adoption: Teams integrate 3-5 bridge SDKs because the 'canonical' solution doesn't exist.
$1B+
Weekly Volume
3-5 SDKs
Avg. Integration
06
The EVM Monoculture is a Feature
Move, FuelVM, and SVM offer technical advantages, but EVM equivalence is the ultimate business development tool.
- Developer Liquidity: ~90% of active smart contract devs target the EVM.
- Capital Efficiency: Deploying on an EVM L2 (Arbitrum, Base) provides instant access to the largest user and capital base.
~90%
EVM Devs
Instant
Access
future-outlook
THE INCREMENTAL PATH
The Endgame: A Truly Scalable Execution Layer
Ethereum's scaling roadmap is a deliberate, multi-phase engineering project that prioritizes security and decentralization over shortcuts.
Full Danksharding is the goal. This final upgrade transforms Ethereum into a massive data availability layer, enabling rollups like Arbitrum and Optimism to post data cheaply and scale transaction throughput to 100k+ TPS without compromising security.
The path is incremental. The roadmap proceeds through Proto-Danksharding (EIP-4844) and Danksharding, because upgrading a live, $400B+ network requires phased testing. Each step, like the blob-carrying transactions of EIP-4844, is a standalone utility that builds the foundation for the next.
Execution scaling is outsourced. Ethereum's core will not process user transactions. Its role shifts to consensus and data, while specialized execution layers—ZK-rollups (zkSync, Starknet) and Optimistic rollups—handle computation. This separation of concerns is the architectural breakthrough.
Evidence: Post-EIP-4844, Arbitrum One's average transaction fee dropped by over 90%, demonstrating the immediate impact of improved data availability. The end-state targets a cost of $0.01 per transaction.
takeaways
THE EVOLUTIONARY PATH
TL;DR for Protocol Architects
Ethereum's scaling roadmap is a deliberate, incremental process where each layer builds on the security of the last.
01
The Data Availability Bottleneck
Rollups are constrained by Ethereum's ~80 KB/s data bandwidth. This creates a hard cap on total network throughput, regardless of execution speed.
- L1 is the bottleneck: Even a 100k TPS rollup is limited by L1's data posting rate.
- The Blob fee market: Introduced with EIP-4844, creating a separate, cheaper resource for rollup data.
- The path forward: Full danksharding aims to increase this capacity to ~1.3 MB/s.
~80 KB/s
Current Limit
16x
Target Gain
02
Layer 2s are Execution Shards
Optimistic Rollups (Arbitrum, Optimism) and ZK-Rollups (zkSync, StarkNet) act as specialized execution environments.
- Security inheritance: They derive finality from Ethereum L1, unlike sidechains.
- Modular specialization: L2s can optimize for specific use cases (e.g., gaming, DeFi) without L1 consensus changes.
- The trade-off: Introduces fragmentation (liquidity, composability) and new trust assumptions for fraud proofs.
10-100x
Cheaper Tx
~7 Days
ORU Challenge
03
The Interoperability Tax
Scaling creates isolated liquidity pools and state. Bridging assets between L2s and L1 reintroduces latency, cost, and security risks.
- Native bridges vs. third-party: Security models vary wildly (canonical vs. LayerZero, Wormhole).
- The composability break: A DeFi transaction spanning Arbitrum and Optimism requires multiple steps and delays.
- Emerging solutions: Shared sequencing networks and unified liquidity layers like Chainlink CCIP aim to abstract this away.
~10-20 mins
Bridge Latency
$5B+
Bridge TVL Risk
04
The Modular Endgame: Celestia & EigenDA
The future is separating execution, settlement, consensus, and data availability (DA). This breaks Ethereum's monolithic scaling model.
- Specialized DA layers: Celestia and EigenDA offer cheaper, high-throughput data posting for rollups.
- Settlement layer competition: Ethereum L1 becomes a high-security settlement hub among many.
- The risk: Over-modularization can increase systemic complexity and create new centralization vectors in DA layers.
-99%
DA Cost
New Stack
Complexity
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall