Vertical scaling is specialization. Horizontal scaling adds parallel general-purpose chains like Arbitrum and Optimism, but vertical scaling builds dedicated, optimized environments for specific applications, such as dYdX's Cosmos app-chain or Aevo's custom L2.
the-modular-blockchain-thesis-explained
Blog
The Future of Scalability is Vertical, Not Just Horizontal
The modular blockchain thesis is not just about splitting data and execution. Its endgame is vertical specialization: purpose-built execution layers that optimize for specific applications, creating a Cambrian explosion of high-performance, sovereign blockchains.
introduction
THE PARADIGM SHIFT
Introduction
Scalability is evolving from horizontal layer-2 expansion to vertical specialization of the execution layer itself.
Horizontal scaling hits diminishing returns. Adding more L2s fragments liquidity and complicates UX, creating a bridge meta-game dominated by protocols like Across and LayerZero. Vertical scaling internalizes this complexity.
The endgame is modular execution. The future stack separates settlement, data availability, and execution, enabling projects like Eclipse and Saga to deploy app-specific rollups with optimized virtual machines for their exact needs.
Evidence: The migration of dYdX from a StarkEx L2 to its own Cosmos chain, which increased throughput by orders of magnitude for its specific orderbook model, proves the vertical thesis.
key-trends
THE ARCHITECTURAL SHIFT
Executive Summary
Horizontal scaling (L2s) solved for throughput but fragmented liquidity and UX. The next wave builds vertically: specialized layers for specific functions.
01
The Problem: The L2 Fragmentation Tax
Users and developers pay a hidden tax for multi-chain existence. Cross-chain bridges and liquidity pools lock up $20B+ in capital that could be productive. Settlement latency and failed transactions degrade UX.
- Cost: ~$5-50 per failed cross-chain swap attempt.
- Complexity: Developers maintain 5-10+ chain deployments.
$20B+
Capital Locked
5-10x
Dev Complexity
02
The Solution: Sovereign Execution Layers (RollApps)
Projects like dYmension and Celestia enable hyper-specialized chains (RollApps) with shared security and data availability. Each app owns its vertical stack.
- Performance: Sub-second finality for app-specific logic.
- Sovereignty: Full control over fee markets and upgrades.
<1s
Finality
~$0.001
Tx Cost
03
The Problem: Monolithic MEV Extraction
General-purpose L1s/L2s create a single, exploitable market for arbitrage. This leads to frontrunning and $500M+ annual value extraction from users, distorting transaction ordering.
- Inefficiency: Value leaks out of the application layer.
- Centralization: Proposer-Builder-Separation (PBS) is a band-aid.
$500M+
Annual Extract
100ms
Arb Window
04
The Solution: Intent-Based Architectures
Protocols like UniswapX, CowSwap, and Across shift from transaction execution to outcome fulfillment. Users state what they want, solvers compete how to achieve it.
- Efficiency: MEV is captured and returned to users as better prices.
- UX: Gasless, cross-chain swaps feel atomic.
+20-50bps
Price Improvement
0
Slippage
05
The Problem: One-Size-Fits-All Data Availability
Paying for full Ethereum calldata is overkill for most apps, creating a fixed cost floor. This limits scalability for high-throughput, low-value data applications (e.g., gaming, social).
- Cost: ~$0.10 per 100 bytes on Ethereum L1 DA.
- Bottleneck: DA is the primary scaling constraint.
$0.10
DA Cost Floor
80%
Blob Cost
06
The Solution: Modular DA & Parallelization
EigenDA, Celestia, and Avail provide cheaper, scalable DA layers. Solana and Monad prove massive gains via parallel execution engines. The stack unbundles.
- Throughput: 100k+ TPS achievable by separating execution, settlement, DA.
- Flexibility: Apps choose security/cost trade-offs.
100k+
Potential TPS
-99%
DA Cost
thesis-statement
THE ARCHITECTURAL SHIFT
The Core Thesis: From Generic Compute to Purpose-Built Engines
Blockchain scaling will be dominated by specialized, vertically integrated stacks, not by faster general-purpose virtual machines.
Generic compute is a bottleneck. The EVM's one-size-fits-all model forces every application to compete for the same resources, creating a zero-sum game for block space and limiting optimization.
Vertical scaling unlocks order-of-magnitude gains. Purpose-built chains like dYdX (Cosmos) and Aevo (Solana SVM) integrate the application, execution environment, and data availability layer, eliminating interoperability overhead and enabling custom fee markets.
The future is application-specific rollups. Frameworks like Eclipse and Sovereign Labs provide the tooling to launch these vertical stacks, turning monolithic L2s like Arbitrum and Optimism into general-purpose settlement layers for specialized rollups.
Evidence: dYdX v4's orderbook processes trades at sub-second finality with zero gas fees for users, a performance profile impossible on a shared EVM rollup.
market-context
THE ARCHITECTURAL DIVIDE
The Current Landscape: Monolithic Silos vs. Modular Commodities
Scalability is shifting from horizontal layer-2 replication to vertical specialization of core blockchain functions.
Monolithic architectures are scaling dead ends. Blockchains like Solana and Avalanche bundle execution, consensus, data availability, and settlement into a single layer, creating performance ceilings and vendor lock-in.
Modular design commoditizes the stack. Projects like Celestia (data availability) and EigenLayer (restaking) decouple functions, allowing specialized networks like Arbitrum and zkSync to optimize for execution.
The future is vertical integration. Winning protocols will own a critical, non-commoditized layer, like StarkWare's prover market or Polygon's zkEVM, not just another generic L2.
Evidence: Celestia's blobspace reduces L2 data costs by 99% versus Ethereum calldata, proving the economic advantage of modular data availability.
THE ARCHITECTURAL DIVIDE
Horizontal vs. Vertical Scaling: A Feature Matrix
A first-principles comparison of scaling paradigms, contrasting the additive, parallel approach of horizontal scaling with the integrative, performance-focused approach of vertical scaling.
| Core Feature / Metric | Horizontal Scaling (e.g., Rollups, Sharding) | Vertical Scaling (e.g., Monolithic L1s, Parallel VMs) | Hybrid Approach (e.g., Solana, Monad, Fuel) |
|---|---|---|---|
Primary Scaling Mechanism | Add more parallel chains/domains (L2s, shards) | Increase single-chain performance (CPU, memory, storage) | Combine high-performance L1 with specialized execution layers |
Developer Experience | Fragmented; requires cross-domain tooling (e.g., LayerZero, Hyperlane) | Unified; single state, atomic composability | Mostly unified with optional app-specific parallelism |
Cross-Domain Latency for Atomic Comps | 2+ block times + bridge delay (5 min - 1 hr) | Sub-second (native to VM) | < 1 sec (within the same state) |
Max Theoretical TPS (Current) | ~100k (sum of all rollups/shard capacity) | ~50k (single-chain peak, e.g., Solana) |
|
State Management | Partitioned & isolated per chain/shard | Global, singular state | Global core state with optional isolated states |
Security Model | Derived from L1 (optimistic/zk proofs) or own validator set | Native; single validator set securing entire state | Native L1 security for core, with options for modular security |
Capital Efficiency | Low; liquidity fragmented, bridges impose cost | High; unified liquidity pool | High core, with optional fragmentation for scale |
Protocol Examples | Ethereum L2s (Arbitrum, Optimism), Polkadot parachains | Solana, Aptos, Sui | Solana (with Firedancer), Monad, Sei V2 |
deep-dive
THE ARCHITECTURE
The Vertical Stack: How App-Specific Rollups Unlock New Primitives
App-specific rollups shift the scalability paradigm from horizontal throughput to vertical specialization, enabling new on-chain primitives impossible on shared L1s or L2s.
App-specific rollups are not just shards. They are sovereign execution environments where the application logic dictates the chain's architecture, from the sequencer to the data availability layer. This vertical integration allows for custom fee tokens, native account abstraction, and specialized state models that generic chains cannot support.
The new primitive is the chain itself. A dApp is no longer a smart contract; it is a protocol with its own block space, economic policy, and security model. This transforms applications like dYdX or Hyperliquid from tenants into landlords, controlling their own scalability and revenue.
Shared sequencers create a trade-off. Using a shared sequencer network like Astria or Espresso provides liquidity and interoperability but sacrifices the ultimate vertical control. The optimal design depends on whether the app values sovereignty over atomic composability with other chains.
Evidence: dYdX v4's migration from StarkEx to a Cosmos-based app-chain demonstrates the performance imperative. The new chain processes orders in a central limit order book model, a primitive economically unviable on Ethereum L1 due to gas costs and latency.
protocol-spotlight
THE FUTURE OF SCALABILITY IS VERTICAL, NOT JUST HORIZONTAL
Protocol Spotlight: The Vertical Vanguard
Horizontal scaling (L2s) spreads load; vertical scaling (app-chains, SVMs) optimizes execution for specific applications, delivering superior performance and sovereignty.
01
The Problem: Generic L2s Are a Compromise
General-purpose rollups force all applications to share the same execution environment, virtual machine, and fee market. This creates a one-size-fits-none bottleneck.
- Performance Ceiling: High-throughput DeFi apps are throttled by social media transactions.
- Sovereignty Loss: Upgrades and governance are dictated by the L2's core devs, not the application.
- Inefficient Pricing: Users subsidize the cost of unrelated, complex operations.
~100ms
Tail Latency
Shared
Fee Market
02
The Solution: Sovereign App-Chains (dYdX, Aevo)
Dedicated blockchains built for a single application, using stacks like Celestia for DA and Cosmos SDK for consensus. This is vertical integration for maximum performance.
- Tailored Execution: Custom VM (e.g., CosmWasm) optimized for order-book matching or perpetual swaps.
- Full Sovereignty: Protocol controls its own roadmap, forkability, and MEV policy.
- Predictable Economics: Fees and revenue accrue directly to the protocol treasury and stakers.
10x+
Throughput
$1B+
Chain TVL
03
The Solution: Parallelized Execution Engines (Monad, Sei)
These are vertically scaled L1s or L2s that optimize at the VM level, using techniques like parallel execution and superscalar pipelining to break the EVM's sequential bottleneck.
- State Access Parallelism: Transactions that don't conflict are processed simultaneously.
- Deterministic Profit: Predictable performance unlocks new DeFi primitive design.
- EVM Compatibility: Maintains developer familiarity while delivering ~10,000 TPS.
10k TPS
Target
~1s
Time to Finality
04
The Problem: Shared Sequencer Centralization
Most rollups rely on a single, centralized sequencer for transaction ordering and execution. This creates a critical point of failure and captures all MEV.
- Censorship Risk: The sequencer can reorder or exclude transactions.
- MEV Extraction: Value that should go to users or the app is captured by the sequencer operator.
- Liveness Dependency: If the sequencer fails, the chain halts.
1
Sequencer
100%
MEV Capture
05
The Solution: Shared Sequencing Networks (Espresso, Astria)
Decentralized networks that provide sequencing-as-a-service for multiple rollups. This is vertical specialization of a critical infrastructure layer.
- Decentralized Trust: A validator set replaces a single operator, reducing censorship risk.
- Cross-Rollup MEV: Enables atomic composability and fairer MEV redistribution across an ecosystem.
- Liveness Guarantees: Redundant nodes ensure the network stays live even if some fail.
Decentralized
Architecture
Atomic
Cross-Rollup
06
The Verdict: Vertical Stacks Win
The endgame is not one chain to rule them all, but a constellation of vertically optimized execution layers—app-chains, parallel VMs, rollup-as-a-service—sitting atop robust horizontal layers for data availability (Celestia, EigenDA) and settlement (Ethereum, Bitcoin). This separation of concerns is how we scale to billions of users.
Modular
Architecture
Billions
User Scale
counter-argument
THE COMPOSABILITY TRAP
Counterpoint: The Liquidity Fragmentation & Composability Trap
Horizontal scaling fragments liquidity and breaks the atomic composability that defines the base layer's value proposition.
Horizontal scaling fragments liquidity. Each new L2 or appchain creates its own isolated liquidity pool, increasing capital inefficiency and slippage for users moving assets between chains like Arbitrum and Optimism.
Atomic composability is destroyed. A single transaction cannot natively interact with contracts on Ethereum, Polygon, and Avalanche. This breaks the core DeFi lego primitive, forcing reliance on slow, trust-minimized bridges like Across or LayerZero.
The UX regresses to Web2. Users manage multiple wallets, gas tokens, and bridge wait times. Protocols like Uniswap deploy separate instances per chain, creating a worse experience than the monolithic Ethereum L1 they aimed to scale.
Evidence: The Total Value Locked (TVL) in cross-chain bridges has stagnated below $20B, a fraction of aggregate L2 TVL, proving capital does not flow freely. Projects like dYdX chose an appchain, sacrificing composability for performance.
risk-analysis
SINGLE-POINT FAILURES
Risk Analysis: What Could Derail Vertical Scaling?
Vertical scaling concentrates power and risk; these are the critical vulnerabilities that could stall or break the paradigm.
01
Sequencer Centralization & Censorship
A single, vertically-integrated sequencer becomes a centralized choke point. This creates systemic risk and regulatory attack surface.
- MEV extraction becomes a monopoly, not a market.
- Censorship resistance is lost, enabling blacklisting.
- Liveness risk: A single operator failure halts the entire chain.
1
Active Sequencer
100%
Control
02
The Shared State Monolith
Vertical scaling bundles execution, settlement, and data availability. A bug or exploit in one layer can cascade, poisoning the entire system.
- Contagion risk: A single VM bug can compromise all apps.
- Upgrade rigidity: Coordinating monolithic upgrades is slow and risky.
- Data availability failure bricks the chain, unlike modular designs with Celestia or EigenDA.
0
Fault Isolation
All
Apps Affected
03
Economic Capture & Stagnation
A vertically-integrated stack creates a closed economic loop. Value accrues to the L1 token, disincentivizing innovation in specialized layers.
- Fee market capture: No competition between execution layers or DA providers.
- Innovation tax: New primitives (like parallel VMs) must be adopted top-down.
- Vendor lock-in for developers, contrasting with the modular 'Lego' model.
1
Fee Sink
0
Market Choice
04
The Interoperability Bottleneck
Vertical chains become isolated performance islands. Bridging assets and state introduces the very latency and trust assumptions they aimed to solve.
- Bridge risk recreated: Native vertical scaling doesn't solve cross-chain communication.
- Liquidity fragmentation: Capital is siloed, reducing capital efficiency.
- Forces reliance on third-party bridges like LayerZero or Axelar, adding complexity.
~20 mins
Bridge Delay
$2B+
Bridge TVL at Risk
future-outlook
THE ARCHITECTURAL SHIFT
Future Outlook: The Verticalized Internet of Sovereign Chains
Scalability will be achieved through specialized, sovereign execution layers rather than a single, generalized L1.
Monolithic L1s are obsolete. They force every application to compete for the same, expensive block space, creating a tax on innovation. The future is a network of verticalized app-chains like dYdX and Aevo, each optimizing for a specific use case.
Horizontal scaling is a commodity. Solutions like Arbitrum and Optimism provide general-purpose capacity, but they lack the customizability and sovereignty required for maximal performance. Vertical scaling provides dedicated throughput and MEV capture.
The interoperability layer is the new bottleneck. The value accrues to the protocols that securely connect these sovereign chains. Projects like LayerZero and Axelar are building the messaging standards, while Across and Stargate handle asset transfers.
Evidence: The migration of dYdX from a shared L2 to its own Cosmos-based chain resulted in a 10x reduction in trading fees and full control over its sequencer revenue, validating the vertical model.
takeaways
THE ARCHITECTURE SHIFT
Key Takeaways
Horizontal scaling (adding more L2s) creates fragmentation. The next wave of performance gains will come from vertical specialization within the stack.
01
The Problem: The L2 Fragmentation Tax
Every new L2 adds to the cross-chain liquidity problem, imposing a ~$100M+ annual tax on users via bridge fees and MEV. This creates a poor UX and stifles composability.\n- Cost: Bridging can add 10-50 bps to every transaction.\n- Latency: Finality delays of 10 mins to 7 days break DeFi apps.\n- Security: Users now trust dozens of bridge multisigs instead of Ethereum.
~$100M+
Annual Tax
10-50 bps
Slippage/Cost
02
The Solution: Sovereign Execution Environments
Vertical scaling means building dedicated, optimized environments for specific applications (e.g., a DEX-specific rollup). This is the thesis behind EigenLayer AVS and Celestia Rollups.\n- Performance: Native app logic enables ~500ms latency and <$0.001 fees.\n- Sovereignty: Apps control their own upgrade path and MEV policy.\n- Composability: Secure, fast messaging via shared settlement (e.g., EigenDA, Near DA).
<$0.001
Tx Fee
~500ms
Latency
03
The Enabler: Intent-Based Abstraction
Users shouldn't need a map of 50 chains. Systems like UniswapX, CowSwap, and Across abstract away chain selection by solving for user intent ('get me the best price').\n- Efficiency: Solvers compete across L2s, capturing >$1B in MEV savings.\n- UX: Single transaction, any chain. The 'chain' becomes an implementation detail.\n- Future: This pattern will extend to gaming, social, and identity.
> $1B
MEV Saved
1-Click
UX
04
The Metric: Economic Throughput
Forget TPS. The real metric is Economic Throughput: the value secured and settled per unit of time and cost. This is what Solana (high vertical integration) and modular stacks (vertical specialization) optimize for.\n- Solana: ~$100B+ in annualized settlement value at ~$0.002 average cost.\n- Modular: Enables parallel economic activity (DeFi, Gaming, Social) without congestion.\n- Result: Capital efficiency improves by orders of magnitude.
$100B+
Settled Value
~$0.002
Avg. Cost
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