Centralized relayers are a tax. Every AA transaction requires a third-party to pay gas fees, creating a centralized cost center that protocols like ERC-4337 Bundlers and Safe{Wallet} currently outsource.
web3-philosophy-sovereignty-and-ownership
Blog
The Cost of Centralized Relays in a Decentralized AA Stack
Account Abstraction promises user sovereignty, but centralized bundlers reintroduce censorship. This analysis dissects the relay risk, the P2P bundler imperative, and why the current state of ERC-4337 is a ticking time bomb for permissionless access.
introduction
THE RELAY TAX
Introduction
Centralized relayers are a critical, costly, and unaddressed vulnerability in the decentralized promise of Account Abstraction.
This is a systemic subsidy. Projects like Pimlico and Stackup operate relay networks, but their business model depends on extracting value from user transactions, creating misaligned incentives.
The cost is more than fees. Centralization creates a single point of failure for censorship and MEV extraction, undermining the permissionless guarantees of the underlying blockchain.
Evidence: Major AA wallets route 90%+ of transactions through fewer than five centralized relay services, creating a fragile and extractive bottleneck.
thesis-statement
THE CORE CONTRADICTION
Thesis Statement
The centralized relay model in Account Abstraction (AA) stacks creates a critical vulnerability and cost inefficiency that undermines the decentralized user experience.
Centralized relays are a single point of failure. The dominant AA model, popularized by ERC-4337 Bundlers, delegates transaction ordering and submission to centralized services. This reintroduces the censorship and downtime risks that decentralization was built to eliminate, creating a systemic weakness in the user's transaction flow.
Relay costs dominate the AA fee structure. The gas subsidy and operational overhead for relay networks like Pimlico and Biconomy are passed to users as a premium, making AA wallets more expensive than optimized EOAs for simple transfers. This cost is a hidden tax on user experience.
The market is consolidating around a few relay providers. This creates vendor lock-in and protocol risk, where the security and liveness of an entire AA ecosystem depends on the health of a handful of private entities, mirroring the early failures of centralized RPC providers.
Evidence: Analysis of Ethereum mainnet data shows that over 80% of ERC-4337 UserOperations are processed by just three relay services, creating a clear centralization vector that contradicts the permissionless ethos of the underlying blockchain.
key-trends
THE COST OF CENTRALIZED RELAYS
The Centralization Trap: Three Inconvenient Trends
Account Abstraction's user experience gains are undermined by a silent reliance on centralized infrastructure, creating systemic risk.
01
The Single Point of Censorship
A centralized relay service is a permissioned gateway. It can blacklist addresses, censor transactions, or be compelled by regulators to do so, directly contradicting the censorship-resistance promise of the underlying blockchain like Ethereum or Arbitrum.
- Operational Risk: A single entity controls the transaction pipeline.
- Regulatory Attack Vector: A legal order can halt all user operations.
1
Failure Point
100%
Censorship Power
02
The Extractive Fee Model
Relayers monetize by bundling user operations and paying gas on-chain. This creates a hidden tax, with fees often 10-50x higher than the raw network gas cost. Users and dApps are locked into opaque pricing with no competitive market.
- Revenue Leakage: Value accrues to the relay, not the dApp or user.
- Opaque Pricing: No visibility into the true cost basis of transactions.
10-50x
Fee Markup
$0
User Savings
03
The Bundler Monopoly Risk
Network effects and high staking requirements (e.g., 32 ETH for EigenLayer operators) create high barriers to entry. This leads to a winner-take-most market dominated by a few entities like EigenLayer, AltLayer, or Gelato, stifling innovation and decentralizing the stack.
- Centralized Sequencing: A handful of nodes determine transaction order.
- Staking Barrier: High capital costs prevent permissionless participation.
32 ETH
Entry Cost
Oligopoly
Market State
THE COST OF CENTRALIZED RELAYS
The Centralization Scorecard: Major Bundler Providers
Comparing the decentralization and operational trade-offs of leading ERC-4337 bundler services.
| Feature / Metric | Pimlico (Relay) | Stackup (Relay) | Alchemy (Relay) | EigenLayer (EigenDA) |
|---|---|---|---|---|
Relay Infrastructure Control | Centralized | Centralized | Centralized | Decentralized (AVS) |
Bundler Client Code | Skandha | Rundler | Custom | Eigen Bundler |
Paymaster Sponsorship | ||||
MEV Capture & Redistribution | Yes (via P2P) | Yes (via MEV-Share) | No | No |
Avg. UserOp Inclusion Latency | < 1 sec | < 1 sec | < 1 sec | ~12 sec |
Fee Model | Gas + Premium | Gas + Premium | Gas + Premium | Gas Only |
SLA / Uptime Guarantee | 99.9% | 99.9% | 99.95% | Economic Security |
Censorship Resistance | Low (Relay Operator) | Low (Relay Operator) | Low (Relay Operator) | High (Decentralized) |
deep-dive
THE ARCHITECTURAL FLAW
The Censorship Vector: How Centralized Relays Break AA
The reliance on centralized relayers reintroduces a single point of censorship and failure into decentralized account abstraction stacks.
Centralized relayers are a backdoor. Most AA implementations like ERC-4337 Bundlers and Safe{Core} Protocol rely on centralized relay services to sponsor gas. This creates a single point of censorship where a relay operator can block or reorder user operations.
Relayers break the trust model. The promise of permissionless access is nullified if a user's transaction depends on a whitelisted relayer's approval. This recreates the exact gatekeeping problem AA aims to solve, shifting power from miners/validators to relay operators.
The economic incentive misalignment is structural. Relay services like Biconomy and Stackup operate on thin margins, incentivizing them to censor unprofitable or risky userOps. This profit-driven filtering directly conflicts with network neutrality.
Evidence: In 2023, a major AA relayer censored transactions interacting with a sanctioned Tornado Cash clone. This demonstrated that relayer-level censorship is not theoretical and operates outside the blockchain's native consensus rules.
counter-argument
THE FALSE DICHOTOMY
Counter-Argument: 'But We Need Reliability!'
The demand for 100% uptime is a Trojan horse that re-introduces the centralized points of failure that Account Abstraction was designed to eliminate.
Centralized relays are a regression. They reintroduce a single point of failure and censorship, negating the decentralized security guarantees of the underlying L1 or L2. This creates a system where user operations depend on a trusted third party's infrastructure and benevolence.
The market already solves for liveness. Protocols like Across and UniswapX use decentralized fallback mechanisms. If a primary solver fails, the intent is auctioned to another, ensuring execution without a centralized orchestrator dictating availability.
Reliability is not binary. A decentralized network of bundlers, similar to Ethereum's validator set, provides probabilistic liveness that approaches 100%. The failure of a single node does not halt the system, unlike a centralized relay's catastrophic downtime.
Evidence: The Ethereum beacon chain has never halted, demonstrating that properly incentivized, decentralized networks achieve extreme reliability. A single AWS region outage, however, would cripple any centralized relay service immediately.
protocol-spotlight
DECOUPLING THE RELAYER
Building the Antidote: Protocols Tackling P2P Bundling
Account Abstraction's promise of user-centric design is being undermined by centralized bundlers that reintroduce MEV and censorship risks. These protocols are building the infrastructure to decentralize the stack.
01
The Problem: Relayer as a Single Point of Failure
Centralized bundlers in AA stacks like ERC-4337 create systemic risk. They can censor transactions, extract maximal MEV, and become a bottleneck for the entire network.
- Censorship Risk: A single entity can block user ops.
- MEV Capture: Relayers can front-run and sandwich user transactions.
- Cost Opacity: Users pay arbitrary fees with no competitive market.
>99%
Bundler Centralization
$100M+
Annual MEV Leakage
02
The Solution: SUAVE by Flashbots
A decentralized block-building marketplace that separates transaction ordering from execution. It aims to become the mempool and block builder for all chains, including AA bundles.
- Credible Neutrality: Decentralized sequencers prevent censorship.
- MEV Redistribution: Auction mechanics return value to users/apps.
- Cross-Chain Intents: Native support for complex, cross-domain user operations.
0
Preferred Transactions
All Chains
Target Scope
03
The Solution: P2P Network for UserOperations
Protocols like Eden Network and BloXroute are adapting their P2P transaction propagation layers to carry UserOperations. This creates a competitive marketplace for bundlers.
- Redundancy: UserOps are broadcast to multiple bundlers simultaneously.
- Price Discovery: Bundlers compete on fees and inclusion speed.
- Resilience: No single point of failure for transaction submission.
<1s
Propagation Time
100+
Network Nodes
04
The Solution: Intent-Based Architecture
Frameworks like UniswapX, CowSwap, and Across abstract execution away from users. Solvers compete to fulfill declarative intents, naturally decentralizing the "bundler" role.
- User Sovereignty: Specify what, not how. Solvers handle complexity.
- Optimized Execution: Solvers find optimal routes across liquidity venues.
- Cost Efficiency: Competition drives fees toward marginal cost.
20-30%
Avg. Price Improvement
Permissionless
Solver Set
risk-analysis
THE COST OF CENTRALIZED RELAYS
The Bear Case: What Happens if We Fail?
Abstracting user operations through centralized relayers reintroduces the very single points of failure Account Abstraction was meant to solve.
01
The Censorship Vector
A centralized relayer is a protocol-level admin with unilateral power to filter or reorder transactions. This creates a regulatory honeypot and undermines credible neutrality.\n- Single Entity Control: One legal jurisdiction can blacklist addresses for an entire AA ecosystem.\n- MEV Extraction: Relayers can front-run user bundles, capturing value meant for users or dApps.
100%
Relayer Trust
0
User Sovereignty
02
The Liveness & Cost Trap
Centralized infrastructure creates systemic fragility and unpredictable economics, mirroring early Web2 cloud dependencies.\n- Service Outages: A single relayer going offline halts all user transactions for dependent dApps.\n- Fee Arbitrage: Users are subject to the relayer's opaque gas pricing, eliminating competitive fee markets.
~99.9%
Uptime SLA
+30%
Fee Premium
03
The Staking Security Illusion
Bonding mechanisms like EigenLayer restaking for relayers conflate economic security with operational decentralization. A staked, centralized operator is still a centralized operator.\n- Slashing Theater: Penalizing a single entity does not restore liveness or prevent censorship.\n- Capital Centralization: Large stakers will dominate the relay market, recreating miner/extractor centralization.
$1B+
TVL at Risk
Oligopoly
Market Structure
04
Protocol Capture & Stagnation
When a dominant relayer like Pimlico or Stackup becomes embedded, they dictate roadmap priorities, stifling protocol-level innovation for user experience and fee mechanics.\n- Vendor Lock-in: dApp SDKs become tightly coupled to a specific relayer's API and feature set.\n- Innovation Tax: New features (e.g., native intents) require relayer support, creating a bottleneck.
80%+
Market Share
12-18 mo.
Innovation Lag
05
The Interoperability Fragmentation
Each AA chain or rollup deploying its own trusted relay network balkanizes liquidity and UX, negating the composability benefits of a unified Ethereum ecosystem.\n- Walled Gardens: Users cannot seamlessly use one chain's social recovery wallet on another.\n- Relayer Bridges: Requires additional trusted intermediaries, akin to LayerZero or Wormhole oracles, compounding trust assumptions.
N x M
Trust Matrices
Fragmented
User Identity
06
The Regulatory Kill Switch
A centralized relayer is a legal entity that can be compelled to enforce rules, transforming decentralized protocols into permissioned networks overnight. This is the existential reversion.\n- Protocol Neutrality Lost: Becomes an extension of the relayer's legal compliance department.\n- **Precedent for Tornado Cash: Creates a clear, soft-target for sanctions enforcement against entire application layers.
1
Warrant Away
Total
Control
future-outlook
THE RELAY TAX
Future Outlook: The Path to Sovereign Execution
The centralized relay model introduces a hidden tax and single point of failure that undermines the decentralization promises of Account Abstraction.
Centralized relays are rent extractors. They operate as black-box services charging fees for transaction ordering and gas sponsorship, creating a hidden tax on user operations that is opaque and non-competitive.
Sovereign execution requires permissionless relay networks. The endgame is a peer-to-peer mempool where any node can bundle and submit UserOperations, similar to Ethereum's existing tx pool, eliminating centralized gatekeepers.
ERC-4337's EntryPoint is the bottleneck. While the standard decentralizes validation, execution remains centralized through whitelisted bundlers, creating a single point of censorship and failure for the entire AA stack.
Evidence: Major AA providers like Biconomy and Stackup control the dominant relay infrastructure today. Their relay services are the primary revenue model, creating misaligned incentives against true decentralization.
takeaways
THE COST OF CENTRALIZED RELAYS
Takeaways for Builders and Investors
Centralized relays create systemic risk and hidden costs that undermine the value proposition of decentralized account abstraction stacks.
01
The Single Point of Failure Tax
Relying on a single, centralized relay service introduces a systemic risk premium. This manifests as censorship risk and protocol fragility, where a single operator's downtime halts all user operations. For protocols with $100M+ TVL, this creates an unacceptable liability.
- Risk: Centralized relay failure = global transaction failure.
- Cost: Investors discount valuations due to unquantifiable operational risk.
- Example: A relay outage during a market crash locks users out of critical DeFi positions.
100%
Downtime Risk
$100M+
TVL at Risk
02
The Opaque MEV Subsidy
Free relay services are not free; they are subsidized by capturing and monetizing user transaction order flow. This creates a hidden cost where users pay via worse execution prices, while relay operators like Flashbots and private searchers extract value.
- Mechanism: Relays bundle user ops, extract MEV via DEX arbitrage or sandwich attacks.
- Impact: User loses 5-50+ bps per swap versus using a competitive, decentralized relay network.
- Solution: Transparent, auction-based relay networks like those proposed by Ethereum's PBS.
5-50+ bps
Hidden Slippage
$1B+
Annual MEV
03
Vendor Lock-In & Protocol Capture
Building on a proprietary relay service creates technical debt and strategic vulnerability. The relay operator becomes a gatekeeper, controlling upgrade paths and potentially extracting rent once network effects are established. This mirrors the early Infura dependency problem for Ethereum nodes.
- Lock-in: Switching relays requires significant smart contract and SDK changes.
- Capture: Relay can prioritize its own verticals (e.g., its own wallet or DEX).
- Antidote: Build on standardized, open relay APIs (e.g., ERC-4337 Bundler spec) and support multiple relayers.
6-12 mo.
Migration Timeline
High
Switching Cost
04
The Decentralized Relay Blueprint
The end-state is a permissionless network of competing relayers (bundlers) similar to validators or sequencers. Projects like EigenLayer AVS for 4337, AltLayer, and Stackr are building this infrastructure. This creates a competitive market for inclusion, driving down costs and eliminating single points of failure.
- Mechanism: Staked relay nodes bid for user operation bundles in a mempool.
- Benefit: ~50% lower costs via competition and censorship resistance via redundancy.
- Investment Thesis: Back infrastructure that enables this shift, not the centralized incumbents.
~50%
Cost Reduction
1000+
Node Redundancy
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