EVM dominance is a bootstrapping hack. Its network effects and developer familiarity created a liquidity flywheel for early L2s like Arbitrum and Optimism, but it imposes a singular execution model on diverse application needs.
crypto-marketing-and-narrative-economics
Blog
The Future of L2s: Beyond the EVM Monoculture
EVM compatibility has been a strategic moat, but is now a technical ceiling. This analysis argues that the next wave of L2 innovation—in performance, cost, and developer experience—will be unlocked by non-EVM virtual machines like Arbitrum Stylus, FuelVM, and zkVMs, breaking the compatibility trade-off.
introduction
THE MONOCULTURE
Introduction
The EVM's dominance is a temporary scaffold, not the final architecture for scalable blockchains.
Monocultures create systemic risk. A single virtual machine target, like the EVM, creates a uniform attack surface and limits innovation in state models and parallel execution, as seen in the design divergence of Solana, Fuel, and Aztec.
The next phase is specialization. High-throughput games require parallel execution engines (Solana SVM, Fuel). Privacy demands zk-native VMs (Aztec, Polygon Miden). The future stack uses the EVM for liquidity access but routes computation to optimal environments.
thesis-statement
THE MONOCULTURE
The Core Argument: The EVM is a Bottleneck
The Ethereum Virtual Machine's dominance is a structural constraint on innovation, forcing all Layer 2s to inherit its fundamental inefficiencies.
The EVM imposes a tax on every L2. Its 256-bit architecture, single-threaded execution, and storage model are inherently inefficient for high-throughput environments. This forces L2s like Arbitrum and Optimism to spend engineering effort on workarounds instead of novel architectures.
Parallel execution is impossible within the EVM's sequential model. This creates a hard performance ceiling that Solana's Sealevel and Sui's Move-based execution already bypass. The EVM's design is the reason L2s struggle to scale beyond a few thousand TPS.
Innovation is bottlenecked by backward compatibility. New L2s like Monad must fork the Geth client and build custom virtual machines to escape the EVM's constraints, proving the monoculture is a liability. The future is multi-VM, not a faster EVM.
key-trends
BEYOND THE EVM MONOCULTURE
The Multi-VM Inflection Point: Three Catalysts
The EVM's dominance is fracturing under the weight of specialized applications, creating a new architectural paradigm for Layer 2s.
01
The Problem: The EVM's Performance Ceiling
The EVM's single-threaded execution and 256-bit word size are a universal tax on all applications, creating a performance bottleneck for high-throughput use cases like gaming and orderbook DEXs.
- Inefficient Compute: Native operations (e.g., hashing, signatures) are 10-100x slower than native execution.
- State Bloat: Every app pays for storage costs designed for the broadest, not most efficient, use case.
~100x
Slower Compute
+90%
Gas Overhead
02
The Solution: Sovereign AppChains & Rollup-As-A-Service
Frameworks like Arbitrum Orbit, OP Stack, and zkStack enable teams to launch application-specific L2s and L3s with custom VMs, turning infrastructure into a configurable product.
- Tailored Execution: Choose a VM (EVM, SVM, MoveVM) optimized for your application logic.
- Modular Security: Inherit security from a parent chain (Ethereum, Celestia, Avail) while controlling your own data availability and sequencer.
50+
Live AppChains
<1 Week
Deployment Time
03
The Catalyst: The Rise of Alternative VM Ecosystems
Mature, high-performance virtual machines with proven developer traction are breaking the EVM's network effects, forcing L2s to become multi-VM hosts.
- Solana Virtual Machine (SVM): Enables parallel execution and sub-second finality on L2s via Eclipse and Nitrogen.
- MoveVM: Provides built-in resource-oriented security for financial primitives, championed by Aptos and Sui.
- CosmWasm: Becomes the standard for interoperability-focused chains within the IBC ecosystem.
$4B+
SVM TVL
10k+ TPS
Theoretical Cap
THE FUTURE OF L2S: BEYOND THE EVM MONOCULTURE
Virtual Machine Landscape: A Builder's Comparison
A feature and performance comparison of major non-EVM execution environments for L2 and L3 development.
| Feature / Metric | Starknet (Cairo VM) | zkSync Era (ZK Stack) | Fuel (FuelVM) | Arbitrum Stylus (WASM) |
|---|---|---|---|---|
Native Language | Cairo | Rust, Solidity, Vyper | Sway (Rust-like) | Rust, C, C++ |
EVM Bytecode Compatibility | ||||
Parallel Transaction Execution | ||||
State Growth Cost Model | Flat fee per TX | Pay-for-byte storage | UTXO-based, minimal state | WASM gas + EVM gas |
Prover Cost (Approx. per TX) | $0.01 - $0.10 | $0.05 - $0.20 | N/A (Optimistic) | N/A (Optimistic) |
Time to Finality (L1) | ~12 hours (ZK proof) | ~1 hour (ZK proof) | ~1 week (fraud proof) | ~1 week (fraud proof) |
Developer Tooling Maturity | Moderate (Dojo, Protostar) | High (Hardhat plugins) | Low (Sway toolchain) | Emerging (Stylus CLI) |
Native Account Abstraction |
deep-dive
THE EXECUTION
Arbitrum Stylus: The Pragmatic Bridge
Stylus is a deterministic, multi-VM environment that extends Arbitrum's EVM dominance without forcing a risky ecosystem fork.
Stylus is a co-processor, not a replacement. It runs Rust, C, and C++ smart contracts in a deterministic WebAssembly (WASM) VM alongside the EVM. This sidesteps the existential risk of abandoning Solidity's $30B+ developer ecosystem, unlike monolithic L1s like Solana or Sui.
The economic model is the unlock. Stylus contracts pay for compute in native machine cycles, not EVM gas. This enables 10-100x cheaper computation for complex operations like ZK-proof verification or game physics, creating new economic niches that pure-EVM chains cannot serve.
Interoperability is native, not bridged. Stylus contracts share memory and call directly into EVM contracts via a secure, low-latency IPC bridge. This contrasts with the fragmented, trust-minimized bridging required by Cosmos IBC or LayerZero, enabling seamless composability between VMs on the same state.
Evidence: The Arbitrum DAO's approval of a 200M ARB grant for Stylus adoption demonstrates the economic priority. Early benchmarks show a Rust-based Uniswap V2 port executing swaps at <5% of the EVM-equivalent gas cost.
protocol-spotlight
THE FUTURE OF L2S
Beyond Stylus: The Contenders
The EVM's dominance is being challenged by new architectures optimized for specific performance and developer trade-offs.
01
The Problem: EVM's Performance Ceiling
The EVM's single-threaded execution and 256-bit word size create inherent bottlenecks for high-frequency trading and complex computations.\n- Inefficient Computation: Gas costs for non-standard ops (e.g., hashing) are prohibitive.\n- Sequential Bottleneck: No native parallelism, capping max TPS.
~15 TPS
EVM Core Limit
10-100x
Gas Cost Premium
02
The Solution: Parallel Execution Engines
Architectures like Aptos Move and Sui Move use a data-centric model for parallel transaction processing, bypassing EVM constraints.\n- State Access Parallelism: Transactions without conflicts execute simultaneously.\n- Predictable Gas: Costs tied to physical resources, not opcode approximations.
160k+ TPS
Theoretical Max
<$0.001
Avg. Tx Cost
03
The Problem: Fragmented Developer Ecosystems
Building cross-chain requires learning multiple virtual machines (EVM, SVM, MoveVM), increasing overhead and security risk.\n- Tooling Duplication: Audits, indexers, and oracles rebuilt per environment.\n- Liquidity Silos: Native assets and composability are VM-locked.
3+ VMs
To Master
$10B+
Siloed TVL
04
The Solution: Universal VMs & Intent-Based UX
Projects like Movement Labs (Move-EVM) and Artela (Aspect Programming) enable EVM compatibility within high-performance VMs. User-centric layers like UniswapX and Across abstract chain selection via intents.\n- EVM-Equivalent Security: Leverage battle-tested tooling on new cores.\n- User Doesn't Choose Chain: Solvers compete for best execution across all L2s.
1 Codebase
Multi-VM Deploy
~2s
Intent Settlement
05
The Problem: Opaque State & Limited Privacy
Fully transparent state and deterministic execution prevent enterprise adoption and sophisticated DeFi primitives.\n- MEV Extraction: Front-running and arbitrage are systemic.\n- No Confidentiality: Transaction amounts and business logic are public.
$1B+
Annual MEV
0%
Native Privacy
06
The Solution: zk-Proofs & Confidential VMs
Aztec and Espresso Systems use zero-knowledge proofs to enable private state and shared sequencing. RISC Zero brings verifiable computation to any code.\n- Selective Disclosure: Prove compliance without revealing data.\n- MEV Resistance: Encrypted mempools and fair ordering.
~10k TPS
zk-Rollup Scale
100%
Data Encrypted
counter-argument
THE COUNTER-ARGUMENT
The Rebuttal: The Network Effect is a Moat
The EVM's developer and user network effect creates a defensible moat that alternative VMs must overcome.
The EVM is a standard. Its dominance is not just about execution speed; it is about a shared developer toolkit of Solidity, Hardhat, and MetaMask. This creates massive switching costs for developers and users, anchoring liquidity and activity to EVM-compatible chains like Arbitrum and Base.
Non-EVM chains face a cold start. A new VM like SVM or Move must bootstrap its own entire ecosystem of wallets, indexers, and dev tools from zero. This bootstrapping cost is the primary barrier, not technical superiority. Parallelized execution is useless without applications.
The moat is economic. The liquidity and user base on major EVM L2s are self-reinforcing. Protocols deploy where the users are, and users go where the apps are. This flywheel is why zkSync and Scroll prioritize EVM-equivalence over novel VM design.
Evidence: Over 95% of all DeFi TVL resides on EVM-compatible chains. The non-EVM leader, Solana, holds less than 3% of cross-chain bridge volume compared to Arbitrum and Optimism, demonstrating the network effect in action.
risk-analysis
THE FUTURE OF L2S: BEYOND THE EVM MONOCULTURE
The Bear Case: What Could Go Wrong?
The rush to scale Ethereum risks creating a fragmented, insecure, and economically unsustainable landscape. Here are the critical failure modes.
01
The Fragmentation Trap
Every new L2 creates its own liquidity pool, user base, and security model, fracturing the network effect. This isn't scaling; it's Balkanization.
- Siloed Liquidity: Moving assets between chains incurs ~$5-50 in bridge fees and ~10-20 minute delays, killing capital efficiency.
- Security Dilution: A $500M TVL L2 secured by a $50M stake is a trivial target compared to Ethereum's $100B+ economic security.
- Developer Fatigue: Supporting 10+ L2 SDKs and RPC endpoints is unsustainable for apps.
10+
Isolated Chains
-90%
Security vs L1
02
The Interoperability Mirage
Cross-chain messaging layers like LayerZero, Axelar, and Wormhole are trusted third parties masquerading as infrastructure. Their security is only as strong as their multisig signers.
- Oracle/Relayer Risk: A 2/3 multisig compromise on a major bridge could lead to $1B+ exploits, as seen with Wormhole and Nomad.
- Intent Complexity: Systems like UniswapX and CowSwap push risk to solvers, creating opaque MEV and failed transaction markets.
- Settlement Finality: "Instant" bridges often rely on optimistic assumptions, with true finality taking hours.
$2.5B+
Bridge Hacks (2022-24)
2/3
Common Multisig
03
The Economic Unsustainability of Rollups
The current L2 business model—selling cheap blockspace—is a race to the bottom. Without sustainable revenue, security and decentralization are the first casualties.
- Sequencer Profit Collapse: With ~$0.01 average tx fees, a top-tier L2 generates only ~$1M/year in sequencer revenue, insufficient to fund decentralized sequencing.
- Data Cost Time Bomb: Relying on Ethereum calldata for DA is a ~90% cost center. True cost savings require offloading to EigenDA or Celestia, trading security for margins.
- Token Utility Crisis: Most L2 tokens are governance tokens with zero cashflow rights, leading to inflationary treasury drains.
~$0.01
Avg TX Fee
90%
Cost is Data
04
The VM Monoculture Stifles Innovation
The dominance of the EVM creates a massive technical debt trap. It optimizes for developer migration, not for solving new problem classes like high-frequency trading or verifiable ML.
- Performance Ceiling: EVM's sequential execution and 256-bit words are inherently inefficient, capping throughput at ~100-500 TPS per rollup even with optimal proving.
- Innovation Lag: Non-EVM chains like Solana, Fuel, and Movement with parallel VMs and native assets are capturing next-gen app developers.
- Wasm vs. EVM War: The rise of Wasm-based execution environments (e.g., Artela) threatens to fragment the developer toolkit further.
~500 TPS
EVM Ceiling
10x
Parallel Speedup
future-outlook
THE FUTURE OF L2S
The 2025 Landscape: A Cambrian Explosion
The EVM's dominance fractures as purpose-built execution environments and parallel VMs create a multi-chain future defined by specialization.
EVM dominance is over. The one-size-fits-all EVM model fails for high-frequency trading, AI agents, and fully on-chain games. New architectures like Fuel's parallel UTXO model and Movement's Move-based VM optimize for specific use cases, trading general compatibility for raw performance.
Interoperability becomes the new moat. The winning L2s will be those with native cross-chain programmability. This is not about simple token bridges like Stargate, but about shared sequencers and intent-based coordination layers that let applications span multiple execution environments seamlessly.
Rollups become a commodity. The value shifts from the rollup stack itself to the application-specific execution layer on top. We see this with Eclipse and Sovereign SDK, which let any chain act as a settlement layer, turning the L2 landscape into a competitive market for VM performance.
takeaways
THE FUTURE OF L2S: BEYOND THE EVM MONOCULTURE
TL;DR for CTOs and Architects
The EVM's dominance is a bootstrapping relic; the next wave of scaling will be defined by purpose-built execution environments and novel state models.
01
The EVM is a Bottleneck, Not a Foundation
EVM compatibility trades performance for developer inertia, creating a ceiling for throughput and innovation. The future is multi-VM.
- Key Benefit: Enables ~10,000 TPS per chain with parallel execution (e.g., Fuel, Eclipse).
- Key Benefit: Unlocks domain-specific languages for DeFi (Move) or gaming (Cairo/Solidity).
10,000+
Theoretical TPS
-90%
State Bloat
02
Modularity Wins: The Rise of Sovereign Rollups & Alt-DA
Monolithic L2s re-centralize control. Sovereign rollups (e.g., Celestia, EigenDA) separate execution, consensus, and data availability, creating resilient, specialized chains.
- Key Benefit: Political sovereignty – upgrade without L1 governance (see Dymension).
- Key Benefit: ~$0.001 per MB data availability costs vs. Ethereum calldata.
100x
Cheaper DA
Sovereign
Governance
03
Intent-Centric Architectures Are Inevitable
Users don't want to manage gas, slippage, or liquidity across 50+ chains. Systems like UniswapX, Across, and Anoma abstract complexity into declarative intents.
- Key Benefit: ~30% better execution via solver competition (see CowSwap).
- Key Benefit: Unified liquidity across L2s, eliminating fragmented pools.
30%+
Better Price
Single Tx
Cross-Chain
04
Parallel Execution is Non-Negotiable
Sequential EVM processing wastes hardware. Parallel engines (Sui, Aptos, Monad, Sei) process independent transactions simultaneously, unlocking real scalability.
- Key Benefit: Sub-second finality for user-facing apps (e.g., gaming, perps).
- Key Benefit: Linear scaling with cores – 10k TPS today, 100k TPS tomorrow.
<1s
Finality
Linear
Core Scaling
05
ZK Everything: The Ultimate Settlement Layer
Optimistic rollups have a 7-day weakness. ZK rollups (zkSync, Starknet, Scroll) provide cryptographic safety from minute one. The endgame is a network of ZK-proven L2s and L3s.
- Key Benefit: Instant, trustless bridging to L1 via validity proofs.
- Key Benefit: Enables native privacy stacks (Aztec, Aleo) without new trust assumptions.
~0s
Withdrawal Time
Validity Proofs
Security
06
The Appchain Thesis is Correct, But Overhyped
Dedicated chains (dYdX, Lyra) optimize for one app but fragment liquidity and security. The winner will be shared sequencers (Espresso, Astria) that offer appchain performance with shared security.
- Key Benefit: MEV capture & redistribution back to the app's economy.
- Key Benefit: Atomic composability across app-specific rollups via shared sequencing.
App-Specific
Optimization
Shared Sec
Security
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